PUBLIC FORUM: THE CONSUMER AND CREDIT SCORING

• creditscoring.com: To not take too much of the nation's time with whining and long-winded stories, asked five succinct questions and even provided them in writing (indeed-- even on this web site prior to the forum). The idea was to get the questions in before the noon break: "... as promised we would like to give the audience a chance to raise some questions and issues so that we can be sure to cover them this afternoon." The questions did not survive lunch. Wasted plane ticket.
• FTC: "I can't think of a subject that more consumers have expressed to us their desire to understand exactly what it is."

• Fair, Isaac: The Wholesome, Brave, Hard-Working Pioneers (introduced as those who "pioneered" -- see PR)
  • Answers the big question from creditscoring.com: What horrible thing will happen if you release the consumer's score to him?
  • Fair, Isaac's presentation: an attempt to either bore you to death, or, if you do pay attention, confuse the issue. Pseudo technical babble about something completely secret-- so there is no way to challenge their statements or methodology, anyway. Here, in a national presentation, they run out of time and skip though their own material like they are ad-libbing a freshman book report. But, just shut up and take what they feed you, America-- it's PROPRIETARY (as they say, often).
  • On Non-Water Cooler Thinking-- Cold, Hard, Serious, Number Crunching, Numbers Don't Lie, Just the Facts, "It's better to count than to guess" statistics:
    • "We saw all of those numbers, so where do they come from? Well, we do not sit around the water cooler at Fair, Isaac and kind of pull them out of our ears, which again may come as a surprise to people... All of the numbers that are in those scoring models are the result of very painstaking, detailed statistical analysis of real data."
    • NON-WATER COOLER MENTALITY: "And so in fact what we've said is, we don't count any inquiries in the 30 days before this particular score is computed. So you can have a hundred inquiries in the month before the score is computed and they won't count at all. And you can have any number of mortgage or any number of auto loans in any --I'm sorry. It's a 14 day period at any time before the score is computer, and they'll only count as one inquiry each."
      So, with all this statistical, analytical, multiple regressionistic, yadda, yadda, yadda calculating going on, why didn't their algorithms spit out some unusual number of days-- like 12, 17, or 26?
  • "For most kinds of credit, 700 or maybe a little bit up in the 700's. Anything above that is considered golden for most kinds of credit. Anything below about 550 is considered awful."
• The national credit bureaus:
  • Answer the big question from creditscoring.com: What horrible thing will happen if you release the consumer's score to him?: "It's an issue I'm not going to be able to address today."
  • A BIG BREAK FOR THE CONSUMER; SCORE ONE FOR THE FTC:
    "MS. TWOHIG: Thanks, Ray. I have one question about something Peter Mahoney said this morning. When he was arguing that the black box was a myth, which, I think, a lot of folks in the room since disagreed with, he said that you can contact each of the three repositories and they will give you what he called a feedback list. I'm not sure if that's the same as a factor list. And I was just wondering if you know anything about that?
    MR. CRESCENZO: The list of factors that are part of --that could be construed as a full list?
    MS. TWOHIG: I assume that's what he was referring to. Yeah. That's what he was referring to.
    MR. CRESCENZO: Well, I have a full list right here.
    MS. TWOHIG: So that is the case, that those factors that are used by the --
    MR. CRESCENZO: The whole list is available.
    MS. TWOHIG: Okay. I just wanted to clarify that. And if anybody called and asked for that, would you mail it out to them?
    MR. CRESCENZO: I would.
    (Laughter.) MS. TWOHIG: Call Ray.
    MR. CRESCENZO: We have been also --one of our other staff members has been working with the scoring development folks to try to expand the comment factor codes --the explanation codes --so that there is an educational message besides just the statement of lack of payment. So it's, you know, more of the evolution process that continues.
    MR. MEDINE: Ray, do you want to make that part of the record of this, so that people can have access to it and have a better understanding of how the process works?
    MR. CRESCENZO: The list?
    MR. MEDINE: Yes.
    MR. CRESCENZO: Yeah, sure."

    Now, go to his web site and try to find the factors "that could be construed" as a full list. Please send email if you find them. Send it if you don't, too.

    And, try the national credit bureaus' web sites:
    Trans Union
    Equifax
    Experian (be sure to "Ask Max")

    Try
    Freddie Mac
    Fannie Mae

    Try Fair, Isaac:
    www.fairisaac.com/servlet/SiteDriver/Content/929
    http://www.fairisaac.com/servlet/SiteDriver/Content/397#4

    Links from this page to the list from these credible sources will be provided when they get around to publishing it on the Internet. We can rank them in chronological order.

    But, don't hold your breath.

    While you're looking, try to find links to the FTC's (or any other) transcript of the Credit Scoring Forum. It may be hard, since "... 3 days prior to the meeting, you could check the following group calendar of events and find no record of this critical public forum: Fair Isaac, Associated Credit Bureau, Equifax, Experian, Trans Union, Fannie May, Fannie Mac."

• MR. GOLDBERG: "Joe Goldberg from the Pennsylvania Attorney General's office and the Bureau of Consumer Protection. I have a very serious concern with what you said, and I also have a very serious concern with the credit bureaus. Nobody wants to take responsibility for the information that you are transferring. What Fair, Isaac is doing and what the other systems do, is making a statement about the character-- the financial character-- of a consumer, and you are conveying that to a third party. It's not purely internal. And there is a reliance on that statement for better or for worse. What you do need to do is, if you're going to take that step, I think you have the responsibility to tell the consumer the basis for the statement about the character, and the effect that changes in their behavior will have on that statement. I disagree that you don't tell them that. You are making a statement and conveying that to a third party. You have an obligation to allow this consumer to act. And by just giving a list of four factors, if that consumer adds to the consumer's detriment, in my opinion that's an unfair or a deceptive trade practice."
  (Applause.)(no response from the opposition)
• Fannie Mae:
  • Myth of the Down Payment: "Because no longer is a low down payment loan by itself the highest risk alone in the bucket. And in fact, it may be --there may be plenty of opportunity to expand into, you know, a greater population of low down payment loans by using credit scores as a compensating factor."
  • Omigod, has it come to Approval By A Computer?:
    • "However, just as with credit scores, the use of Desktop Underwriter is meant to compliment the underwriter's judgment. It does not replace the underwriter's judgment."
    • Then on the same page: "An approved recommendation, which means that the loan meets Fannie Mae's credit underwriting guidelines and does not need further review by an underwriter."

      Eyewitness account: read from a prepared script.

• Freddie Mac - "High Energy" and Humor (or Comedy):

  "Credit Scoring in the Mortgage Industry: Myths and Facts"

  1. "MYTH: Only One Type of Credit Score is Used in Mortgage Lending"
     
  2. "MYTH: Credit scores make lending decisions"
  3. "MYTH: Credit scoring models are a 'black box' and we don't know how they work"

     "You can contact each of the three repositories, and they will give you a list of the feedback messages --and there's a long laundry list of them --that each of the three repositories use when they send a credit score."

  4. "MYTH: Credit Scores are not based upon a representative population sample"
     • "... we have confirmed that generic credit scores --the stuff that's in FICO scores that we've recommended --are based upon a representative sample based on some of the stuff that Pete showed you, which is that there is a 6.7 percent on minority area representation versus an 8.3 percent of the overall population."
     • On an Experian page: "Developed from a representative national sample of roughly 1 million Experian consumer credit histories, the Experian/Fair, Isaac Model:... "
  5. "MYTH: Credit scoring has not been independently validated"

     "That is not an accurate statement. The Federal Reserve Board --actually Bob Avery is here, and you can ask him about this. They obtained some credit scores. They tested it against low income populations, as well as a broad sample of populations. Their findings were published in the July 1996 Federal Reserve Bulletin. I'm still looking for my autographed copy."

     Looking for any copy.
     Freddie Mac quotes the Federal Reserve.

     Checking with the Federal Reserve on that:

     • MS. WELSH: Hi. Kristy Welsh, K-R-I-S-T-Y, W-E-L-S-H, and I'm with creditinforcenter.com. And I had a question about what Mr. Cook said, that the Federal Reserve Board regulates or somehow evaluates the FICO scoring. How is this done? How is it tested? Are we going through a statistical analysis? Do we take case A, B, C and B and run them through the scoring? Hey, we know for sure that it matches what it should match. And if so --I mean, you said that this was done. Is this published anywhere? Can we get a copy of it?
       MR. COOK: Well, I may have mis-spoke. I don't think I said that we evaluate credit scoring systems per se. I will say that HUD has currently embarked on a study of the scoring systems of the GSE's --Fannie and Freddie. No. I do --I can tell you that in 1996 --and, Bob can correct me if I'm wrong, if he's still here --the Fed did undertake an analysis of credit scoring systems in the sense of trying to determine whether or not there were solid correlations between low scores and poor performance. And generally they found that there were. That credit scoring systems do tend to be predictive of behavior. But that leaves a lot of room, I think, for issues today.
       MS. WELSH: Right. So they haven't been evaluated for bias or discrimination or any of these kinds of things?
       MR. COOK: No. This was not an exhaustive study of each factor in a given scoring system and a relationship. I guess one of the questions that I didn't quite follow up on with Pete, you stated that the statistics are available for your clients to understand and appreciate the various risks associated with each of the factors, so that if they're called upon to justify the inclusion of a particular factor, it's predictiveness can be quantified for them based on the statistics that you've done. But let me ask you this, sort of a further question. Recognizing --sort of taking off on Debby's approach --that we're talking perhaps more than just the legal requirements here. Recognizing that probably you don't have a legal obligation to do the rest of the analysis, to do the third stage of a --what did we call it --disparate impact analysis, that while there may be a justification for using something like finance companies --a number of finance companies --is that the least --does it produce the lowest impact, or is there another set of factors that you could have used that would produce a lesser impact?
     • Federal Reserve: "Credit scoring models are proprietary and consequently public scrutiny is precluded. Suppliers such as Fair, Isaac & Co. provide substantial data to support the validity of their scores but do not publicly release information about how the formulas work."

  6. "MYTH: Credit scoring is unfair to low and moderate-income borrowers"
  7. "MYTH: Credit scoring is unfair to minority borrowers"
  8. "MYTH: Credit scores are always predictive for every single person"
  9. "MYTH: People without credit scores cannot obtain a mortgage"
  10. "MYTH: Credit scoring reduces the availability of low-cost mortgage credit"

• Kirk Willison, Senior Vice President and Director of Government Affairs for Countrywide Home Loans:
  • "... And after listening to a lot of discussion today, I've become convinced that Congress was wrong, in that Congress should not have said that consumers really aren't entitled to their scores... "
  • SCORE ANOTHER ONE FOR THE FTC. The worst appears to be true-- it is contractual:
    "MS. TWOHIG: Kirk, I just want to make sure I understood one thing. Is it the case, though, that you don't give consumers their actual score when you explain to them the reasons why they might be denied?
    MR. WILLISON: Again, the agreement that we have with Fair, Isaac, is that we are not to release that score.
    MS. TWOHIG: Okay.
    MR. WILLISON: Our hands are tied."
• Marcia Griffin, founder and President of Home Free USA:
  • "They need to know their credit score, and they need to know it early... "
  • "... the first secret is that the process has changed. And the second secret is to get your credit score. Get your credit report immediately, not from the credit reporting agencies, because when you write, they don't give you your credit score. They need to really get their credit report with a credit score, so that they can understand where they are... "

TRANSCRIPT STARTS HERE

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 1.
1 Page 2 3
2
1
2 P R O C E E D I N G S
3 -----
4 (9: 02 a. m.)
5 MR. MEDINE: Good morning, and thank you all for
6 coming. This is a tremendous turn out in the middle of
7 the summer on a very, very important issue to consumers.
8 And I would like to begin by introducing the Director of
9 the Bureau of Consumer Protection of the Federal Trade
10 Commission, Jodie Bernstein.
11 MS. BERNSTEIN: Thank you, David. Thank you for
12 the introduction. And I guess they thought I needed it,
13 because I see they've made two name tags for me, which
14 either means that I won't remember my name, or we need a
15 special introduction.
16 I'm delighted to be here this morning to welcome
17 all of you to what's going to be a very important forum.
18 And it's particularly important, as David said, that so
19 many of you were able to get here. I know a lot of you
20 travelled from out of town. And we particularly
21 appreciate it, because the diversity here in the room of
22 views and expressions is going to be critical to us.
23 I can't think of a subject that more consumers

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 2.
2 Page 3 4
3
1 have expressed to us their desire to understand exactly
2 what it is. As someone said, when we get a call from
3 consumers --and we do get lots of calls about a variety
4 of subjects --it's as if there's a black box somewhere
5 and people --somebody somewhere is dropping information
6 into the black box and a score emerges. It's just almost
7 that mysterious to people, they tell us.
8 And then they want to know --and it really does
9 sound like the SATs --how can I improve my score. I
10 know that we won't --we won't produce a list that we can
11 tell consumers how to do that. It's much more complex
12 than that. But I do hope that with this group of experts
13 and the discussion we hope will occur today, that we will
14 all emerge with a good deal more knowledge and
15 understanding of what credit scoring is about. We know
16 it can be a huge benefit to consumers, and we will to be
17 able principally to be able to explain it to them and
18 ensure that they really do understand it and that they're
19 fairly treated.
20 So with the excellent panelists we have today,
21 again, I want to thank all of you for coming and for
22 agreeing to guide our discussion. I could not close
23 without noting that we have true experts to work with us

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 3.
3 Page 4 5
4
1 this morning. And I believe our first speaker this
2 morning is the gentleman sitting to my right, am I right?
3 MR. MEDINE: He is. I wanted to make a
4 housekeeping remarks first. I'm David Medine, the
5 Associate Director for Financial Practices here at the
6 FTC. This forum, first of all, would not have been
7 possible without the two people to my left, Kellie
8 Cosgrove and Peggy Twohig. And I want to congratulate
9 them for all their tremendous effort in putting this
10 together.
11 (Applause.)
12 MR. MEDINE: Cynthia Lamb also was a tremendous
13 assistant in organizing this event, and she's sitting in
14 the back.
15 A couple of just logistical points before we get
16 started. First, there will be a transcript of today's
17 meeting, and as per Commission practice, a copy of the
18 transcript will be available on our web site at
19 www. ftc. gov once it's been prepared.
20 As Jodie mentioned, the purpose of today's forum
21 is to discuss a wide range of issues associated with
22 credit scoring. We want this to be a discussion in which
23 everyone has an opportunity to participate, although the

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 4.
4 Page 5 6
5
1 first couple of sessions will be laying out some of the
2 background. But as the morning progresses, we're going
3 to encourage people to ask questions and to make comments
4 as we proceed, from the audience as well as from the
5 panel members.
6 And for those who are in the overflow rooms,
7 feel free to come by and join us with questions and
8 comments, although we may ask you to return to the
9 overflow rooms, given space considerations. But we don't
10 want people not in Room 432 to be denied the opportunity
11 to participate in today's discussion.
12 If you have any questions or need directions,
13 keep an eye out for the folks with yellow tags on them.
14 They're Commission staff, and they will be happy to
15 direct you to rest rooms, which are just outside and to
16 the left, or to our wonderful Top of the Trade, if you
17 are hungry and need a snack later on, on the seventh
18 floor.
19 It's my pleasure to start the morning off with a
20 good friend of ours, Pete McCorkell, to give us an
21 overview of credit scoring. Pete is the Senior Vice
22 President and General Counsel of Fair, Isaac and Company.
23 Jodie talked about the black box. Well, Fair, Isaac

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 5.
5 Page 6 7
6
1 makes the box, and so this will be a wonderful
2 opportunity for us to get a peak inside the box and get a
3 better sense of how the scores --credit scores are
4 prepared.
5 I think we all discovered that credit scoring
6 had entered the mainstream when in an episode of Murder
7 One a couple of years ago, someone talked about their
8 FICO score as if everyone knew what that meant. And so I
9 think credit scoring moved into the mainstream at that
10 point.
11 Fair, Isaac, of course Pete's company, pioneered
12 the commercial development of empirically derived,
13 predictive models for the credit industry and has really
14 popularized and expanded their use. They are located in
15 California, and they have subsidiaries dealing with the
16 whole issue of decisions using scoring methodology.
17 Pete supervises Fair, Isaac's legal affairs. He
18 joined them in 1987, and has provided advice and
19 assistance to credit grantors in connection with consumer
20 litigation and regulatory proceedings. He is the
21 company's primary liaison with the federal credit
22 regulatory and enforcement agencies, including the FTC,
23 and has conducted seminars for us and other agencies to

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 6.
6 Page 7 8
7
1 educate us about the scoring. And today he'll educate
2 you about the scoring as well.
3 Pete?
4 MR. MCCORKELL: Thanks, David. When I came in,
5 I used a line from Butch Cassidy and the Sundance Kid. I
6 said, can I move. So I'm going to get up and move.
7 Okay. I'm going to sort of approach credit
8 scoring a little bit like peeling back an onion. We've
9 got a lot to cover in the next hour, and so kind of put
10 on your seat belts and try to stay with me.
11 We're going to cover a number of topics, and
12 really in some sense the most important part of what I've
13 got to say is really bringing everybody's awareness about
14 what the nature of the credit decision is. Because an
15 awful lot of the alleged shortcomings of credit scoring
16 turn out to be problems that are inherent in the nature
17 of the credit decision and applied other ways of making
18 credit decisions.
19 I'm also going to talk about sort of the
20 differences between statistically based decisions and
21 judgmental decisions. I'm going to then go into a little
22 bit, hopefully in a very nontechnical way, about how
23 credit scoring systems are developed. I'm going to talk

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 7.
7 Page 8 9
8
1 about different types of credit scoring systems. I'll
2 focus a little bit on the credit bureau scoring systems,
3 because I think that's probably what most people have in
4 mind when they hear the term credit scoring or FICO
5 scores. And then finally, if there's enough time, touch
6 very briefly on some legal and public policy issues
7 related to credit scoring.
8 Actually, one more. Casey Stengel actually had
9 it about right. Making predictions about the future can
10 be kind of dangerous. But in fact, that's what credit
11 grantors have to do. They have to make predictions about
12 how borrowers will behalf, how borrowers will repay that
13 obligation. And when you come right down to it, there is
14 no way to predict with certainty how any individual
15 borrower is going to behalf, how they're going to repay.
16 But what you can do, is that you can predict how
17 groups of borrowers will repay with reasonable certainty.
18 You can predict that if I have a thousand customers with
19 these characteristics, 900 of them, or 950 of them, or
20 990 of them will repay in a satisfactory manner and the
21 others won't. But nobody can predict that this
22 particular borrower will or will not repay in a
23 satisfactory manner.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 8.
8 Page 9 10
9
1 And so what we're talking about is not
2 predicting with certainty the behavior of an individual,
3 but really trying to estimate the risk that different
4 borrowers pose. And risk by its very nature is assuming
5 that you're looking at a group of people, because
6 ultimately every borrower is either going to behave in a
7 satisfactory manner or they're not. They're either going
8 to repay in a reasonably satisfactory manner, or they're
9 going to go delinquent. Maybe go all the way into charge
10 off.
11 So at the end of the day with any particular
12 borrower, you can say, yeah, this was an absolutely great
13 borrower, or this was a terrible borrower. I should
14 never have extended that loan. But when you're making
15 that decision, you can't say for sure that this borrower
16 will or will not repay. And that really gets into kind
17 of the meat of credit scoring and what it's all about.
18 Basically there are fundamentally two ways of
19 making credit decisions. The sort of old fashion
20 judgmental type of decisions, where somebody takes a look
21 at whatever information is available, and sits down and
22 scratches his or her head, or pats his or her stomach or
23 whatever they do to think great thoughts, and comes to a

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 9.
9 Page 10 11
10
1 conclusion that this is a loan that I'm willing to make.
2 The other type of decision process is --and in
3 many cases the two are actually used jointly in any given
4 case. But fundamentally the other type is using the
5 statistical method of making decisions, and that broadly
6 is what I'm going to refer to as credit scoring.
7 Now, a little bit later on I'm going to get into
8 some terminology. In the mortgage industry, at least,
9 credit scoring has come to have, for at least some
10 people, a somewhat more narrower definition. But at
11 least for right now, I'm using credit scoring in kind of
12 the broadest sense of any kind of statistically based
13 credit decision system.
14 So what are the differences between those two
15 types of decisions? Well, actually what I really want to
16 talk about first is what are the similarities. Both
17 methods assume that the future is in some large measure
18 going to resemble the past. Both methods, you're taking
19 a look at the applicant that you've got to make a
20 decision on today, and comparing that applicant with
21 qualities or characteristics of borrowers that you've had
22 experience with in the past, and making the assumption
23 that if today's applicant looks like the people who

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 10.
10 Page 11 12
11
1 generally repaid in a satisfactory manner in the past,
2 then this applicant is likely to repay in a satisfactory
3 manner.
4 The days of Mr. Bailey and the Savings and Loan
5 from It's a Wonderful Life are long gone. Very few of us
6 are known personally to loan officers. Most credit today
7 is extended by mail, sometimes over the phone and
8 increasingly over the Internet. It's probably been 15
9 years since I've had an up front personal encounter with
10 a loan officer, and I suspect that the same is true for
11 most consumers.
12 Today you just don't have that friendly
13 interaction across the desk, and even when you do, you're
14 likely to have --you're not going to have it with Mr.
15 Bailey who knows everybody in town and has known you and
16 your parents and your grandparents. You're going to have
17 it with somebody that doesn't know you from a hole in the
18 wall.
19 And so while your own past history is certainly
20 an important part of the credit decision, in many ways
21 whether somebody is making judgmental or statistical
22 decisions, they're comparing what they know about you to
23 what they know about other applicants and how those other

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 11.
11 Page 12 13
12
1 applicants behaved as borrowers.
2 And then finally, it gets back to the idea of
3 trying to grant credit to acceptable risks. Now, a lot
4 of people making judgmental decisions may sort of fool
5 themselves into thinking well, I know for sure this
6 borrower won't repay, or I know that if I extended credit
7 to this guy, he would not repay. But in fact, you just
8 can't make those individual decisions with any degree of
9 certainty, so in either case you're really trying to hit
10 that acceptable degree of risk.
11 Okay. Now, what are the differences? Well,
12 credit scoring really defines that degree of risk. It
13 allows you to come up with a numerical measure of the
14 degree of risk that any borrower presents. It lets you
15 rank order those borrowers with respect to one another.
16 And I'll get into that a little bit later.
17 It explicitly lets you make those decisions
18 based on a degree of risk. A lender can say, I can make
19 money if I extend credit on this product to people where
20 10 of them will repay for every one that won't, and
21 therefore I have to have at least a 10 to one odds of
22 repayment in order to make money. So I'll extend credit
23 to everybody whose odds are 10 to one or better. If

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 12.
12 Page 13 14
13
1 they're worse than 10 to one, I'm not going to extend
2 credit to those people.
3 It lets you track how those borrowers perform
4 based on the criteria that you're using in the credit
5 scoring system. It lets you make adjustments in your
6 policy. If somebody is making judgmental decisions and
7 the boss comes in and says, our losses are too high,
8 we've got to tighten up, every credit officer is going to
9 have a different idea of what tighten up means. Or if
10 the boss says, gee, our market share is slipping, we're
11 losing business to the competition, we've got to loosen
12 up, every loan officer has to figure out what loosen up
13 really means.
14 With credit scoring somebody could come in and
15 say, we need to tighten up. Let's raise the cutoff by
16 five points. And because of the statistical work that
17 goes into the system, the management knows that if they
18 raise the cutoff by five points, they'll reduce the
19 acceptance rate by X amount and also reduce the
20 delinquency rate by Y amount.
21 And then finally, credit scoring allows the
22 automation of a lot of credit decisions, which I think
23 some of the other folks are going to talk about, but

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 13.
13 Page 14 15
14
1 which has really been a big part of the acceptance of
2 credit scoring. Certainly in small business lending, for
3 example, probably more people have gone to credit scoring
4 for the benefits of automation.
5 In small business lending, for example, the
6 typical statistics we hear are that it takes 10 to 12
7 hours of loan officer time to process a small business
8 loan judgmentally. Using credit scoring and automated
9 underwriting, it can be done in 15 to 30 minutes. Now,
10 if you can take 10 or 11 and a half hours out of the
11 process --and that's not data entry time. That's fairly
12 high paid time. If you take that much time and salary
13 expense out of the process of making a small business
14 loan, you can make a lot more small business loans,
15 especially small dollar small business loans, than you
16 could make if you were doing it judgmentally.
17 One way to sort of think about the differences
18 is that probably in most cases, even somebody making a
19 judgmental decision has a list of factors that they're
20 going to look at when they are reevaluating an applicant.
21 And, again, to just get some lingo in here, what we call
22 characteristics are really the different types of
23 information that are looked at the decision process.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 14.
14 Page 15 16
15
1 Another way that you can think about
2 characteristics is that you might think that each
3 characteristic implies a question that you could ask and
4 then get the answer, either from a credit application or
5 from a credit report.
6 Next slide, please.
7 And so in this case, at the top of the page
8 we've got some information that you might get from a
9 typical credit application, and then down towards the
10 bottom below the blanks are a number of characteristics
11 that you might get off of a credit bureau report. And
12 somebody making a judgmental decision is going to be able
13 to go through that list of characteristics, and probably
14 about the best that they're going to do is be able to say
15 well, this guy looks pretty good on this, but not so good
16 on that.
17 In this case, time at present address looks
18 good, and time on job looks good. Oh, but this guy is a
19 renter and not a homeowner, so that's not so good. Debt
20 ratio is good. He's got a good bank reference. The N/ A
21 after age doesn't mean that there's no age. It means
22 that legally age is not supposed to be used in judgmental
23 credit decisions because of ECOA in Rate B. But they do

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 15.
15 Page 16 17
16
1 allow it to be used in scored decisions.
2 And so you just go through the list like that,
3 and at the end of the process judgmentally you say, oh,
4 this is an acceptable borrower, and that again implies
5 that this is a borrower who poses an acceptable level of
6 risk. You couldn't begin to site the odds of that
7 borrower if you evaluate that borrower using judgmental
8 processes.
9 With credit scoring --and again here is another
10 thing to keep in mind. You're probably going to go
11 through very much the same list of factors that somebody
12 would go through judgmentally. And that's not just sort
13 of, you know, conjecture on our part. One of the things
14 that Fair, Isaac does, is when we build a credit scoring
15 system for somebody that hasn't used credit scoring
16 before, we also build a statistical model of their prior
17 judgmental decision process.
18 And so we can actually find out how --what
19 factors at least appear to be important in that
20 judgmental decision process, and almost invariably will
21 find that very much the same factors are being considered
22 --at least apparently considered --in the judgmental
23 process as wind up being considered in a credit scoring

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 16.
16 Page 17 18
17
1 system.
2 And so the list of factors, the types of
3 information that are considered, are very similar.
4 They're not --again, other than age, which for the legal
5 reasons can't be --or least maybe more accurately, isn't
6 supposed to be used in judgmental decisions. Other than
7 age, the types of information considered are going to be
8 pretty much the same.
9 But with a credit scoring system, for each of
10 those questions that are implied by the characteristics,
11 each possible answer to that question is going to have a
12 number of points assigned to it. And so you could add up
13 all those points and you get the overall score. In this
14 case, you get the same decision, that is an acceptable
15 borrower. In other words, a borrower that poses an
16 acceptable level of risk.
17 But you get one very important additional piece
18 of information, and that is, you get told this borrower
19 is 11 to one. In my example earlier, I said we could
20 afford to extend credit to people at 10 to one odds.
21 This guy is 11 to one, so we'll take him.
22 Next slide, please. This is a sample --and let
23 me stress a sample --of what a credit --the guts of a

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 17.
17 Page 18 19
18
1 credit scoring system, the credit scorecard itself, might
2 look like. It is not an actual credit scoring scorecard,
3 so don't go home and try to score yourself on this and
4 get into a major tiz about your score.
5 I'll just use it to illustrate a couple of
6 points here. First of all, in this case we've got the
7 characteristics down the left hand column. Again, those
8 questions --those types of information. And now across
9 in each row next to the characteristics, we've got the
10 different attributes. The different possible answers for
11 each of those questions, and you'll see that each of them
12 have a different number of points assigned.
13 A couple of things that I want to point out on
14 here, you'll see --and again above the jagged line are
15 pieces of information typically that you get from a
16 credit application, and below the line are
17 characteristics that you would typically get from a
18 credit report.
19 For all of those application characteristics,
20 and actually implicitly also for the bureau
21 characteristics, at the very far right of each column --
22 each row --there is a box labelled NI. That stands for
23 no inform, which is Fair, Isaac speak for no information

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 18.
18 Page 19 20
19
1 value. That may be different than the questions not
2 answered.
3 If you look at the fifth line down, Department
4 Store/ Major Charge Card, you'll see there is a box for no
5 answer. Sometimes when a question is asked on an
6 application and the applicant leaves that blank, the fact
7 that they left it blank tells you something. On the
8 other hand, in a lot of cases credit lenders change
9 applications, or for one reason or another a particular
10 piece of information may not be available for a given
11 individual.
12 We have figured out that it would not be fair to
13 assume that that means that person should get zero points
14 for that piece of information. What they get is what we
15 call the no inform points, which if you go to the next
16 slide, is really can of the neutral value. No inform
17 means that there is no information value to say that this
18 applicant is either better or worse than the average
19 applicant based on that particular piece of information.
20 So if you have a characteristic like age, where
21 you've got, you know, kind of a nice, smooth numerical
22 scale in most cases, there is going to be some age value
23 that's exactly equal to the no inform points. For a

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 19.
19 Page 20 21
20
1 given applicant, not because the applicant refused to
2 answer something, but because you didn't ask the question
3 on an application, if you're missing a piece of
4 information, it will go to the no inform points.
5 In other words, it says we don't have a piece of
6 information on this factor that tells us that this
7 applicant is likely to be risky or less risky than the
8 average applicant who walks through the door. So that's
9 what the NI or no inform points are about.
10 Okay. We saw all of those numbers, so where do
11 they come from? Well, we do not sit around the water
12 cooler at Fair, Isaac and kind of pull them out of our
13 ears, which again may come as a surprise to people. I
14 read something just last week that talked about the
15 scoring model assumes that. Well, scoring models don't
16 assume anything. All of the numbers that are in those
17 scoring models are the result of very painstaking,
18 detailed statistical analysis of real data.
19 The first thing that we have to do in developing
20 a scoring model is to get a sample of borrowers where we
21 can say, this borrower's performance is known to the
22 extent where I'm willing to say that if I knew when I
23 made the decision what I know now about how that borrower

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 20.
20 Page 21 22
21
1 has actually performed, I could say either I would make
2 the same decision to extend them credit, or if I had
3 known then what I know now, I wouldn't have extended
4 credit.
5 And that's what we mean by goods and bads here.
6 I'll use that term again probably throughout the rest of
7 my hour here. That's not a moral judgment or anything
8 else. In some senses, the good and bad refers to the
9 credit grantor's own decision. Did I make a good
10 decision in extending credit to that borrower, or did I
11 make a bad decision. Or in a sense in kind of the way we
12 define it, if, again, knowing how that borrower actually
13 performed you would still extend them credit, then that
14 counts as a good. If you knew how they would perform and
15 you wish you hadn't extended them credit, then that
16 counts as a bad.
17 For the people that you've been accepting under
18 --and let's assume this is somebody that's been using
19 judgmental decisions. For the people that have been
20 accepted before, of course it's very easy to figure out
21 who the goods and the bads are. You just wait a couple
22 of years, depending on the type of credit product, and
23 you see how they perform. You ask yourself, am I glad I

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 21.
21 Page 22 23
22
1 extended credit or not.
2 And so you can count the known goods and bads
3 out of that Accepts on the right hand side there. Again,
4 just to use nice round numbers here, we're assuming we
5 start out with 10,000 people. Under the old decision
6 process, 7,000 were accepted and 3,000 were turned down.
7 For the 7,000 Accepts, figuring out the goods and bads is
8 just a matter of counting based on how they actually
9 performed.
10 But in building a scoring system, we don't want
11 to ignore these 3,000 rejects. The example I gave of
12 somebody saying I've got to have 10 to one odds to make
13 money means that somehow if you could improve your
14 decision process, if you could make sharper decisions
15 about individual risk level, for every 11 people you turn
16 --if you were turning down people that were below 10 to
17 one, for every 10 people you turn down, in certain groups
18 at least, as many as nine of them would have turned out
19 to be good customers if you would have accepted them.
20 And, again, that's an important concept to grasp
21 for lots of types of --in fact, virtually any kind of
22 credit product. To set a cutoff score for economic
23 efficiency, the credit grantor is going to wind up

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 22.
22 Page 23 24
23
1 setting the cutoff score at a point where more of the
2 people that are turned down would have been good, and
3 sometimes by a very large majority. Sometimes it may be
4 10 to one. It may be even more than ten to one. Odds
5 are the point where it becomes unprofitable, because the
6 loss for one bad customer is a lot greater than the
7 profit from any single good customer.
8 And so when somebody says, you know, our cutoff
9 is 415, that doesn't mean that everybody who scores 414
10 and lower is going to go delinquent, and in fact at 414,
11 an awful lot of those people would probably have
12 performed just fine if you had accepted them. But it's
13 not economical given the facts of that particular
14 portfolio to extend credit to those people whose odds are
15 worse than that.
16 We don't want to ignore them, because in fact we
17 know that there is good business in those prior rejects.
18 And so what we go through is a process that has a number
19 of names. The one that I think is most descriptive is
20 Reject Inference. For those of you who paid attention to
21 your high school math classes, you will remember a
22 process called extrapolation, where you could infer where
23 a point was on a line where you didn't actually have any

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 23.
23 Page 24 25
24
1 sample points. It went beyond the ends of the line when
2 you're putting stuff on graph paper.
3 Remember when you were a junior or sophomore in
4 high school, you were putting stuff on graph paper, and
5 if you only had samples, they were sort of in the middle
6 of your graph. You could nevertheless extrapolate to
7 where some points would be outside where you had actual
8 samples.
9 Well, Reject Inference, you can think of that as
10 kind of a multi dimensional form of extrapolation, where
11 again we look at the rejects. We see how they compared
12 on all of the relevant characteristics to the Accepts,
13 and we infer from that, we estimate from that, we
14 extrapolate from that, how they would have performed.
15 And indeed we see here, that of those rejects,
16 the odds were about nine to one. They're almost --
17 they're about nine times as many inferred goods as there
18 are inferred bads in the Rejects. So to build a scoring
19 system, we fold them back in with the known goods and the
20 bads, and, again, hopefully the prior decision process
21 made some sense.
22 And so the odds from actual performance of the
23 people that were accepted, the overall odds for that

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 24.
24 Page 25 26
25
1 group --not the odds at cutoff. But the odds for that
2 group as a whole are 24 to one. The odds for the Rejects
3 are 8.7 to one. And again, I sort of use these numbers
4 because they will work out well for my next example. The
5 overall odds for that population turn out to be 16 to
6 one.
7 So if all that I know is that somebody has
8 walked in the door in this population, I can say, well,
9 if I extended credit to everybody that walks in the door,
10 for every 16 people that paid in a satisfactory way, one
11 of them wouldn't. And that's actually pretty --that's
12 pretty high odds. More typically, in most populations
13 we're going to see something --in the population odds,
14 we're going to see something on the order of ten to one,
15 eight to one and something like that.
16 But as I said, that 16 to one number happens to
17 work well for my next slide and actually several slides
18 after this. So we've now sort of divided the world into
19 the goods and the bads. There aren't any uglies, just
20 goods and bads. So now we continue the process of a very
21 careful analysis of actual data, and we see how people --
22 look at each characteristic, and we look then at each
23 attribute of each characteristic, and we see how those

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 25.
25 Page 26 27
26
1 people perform.
2 And my first example here is age. And we'll see
3 that for people under 30, that group had 10 percent of
4 the goods, but 40 percent of the bads. And so the
5 information value for that attribute is one to four. And
6 again, these numbers here are in percentages, not in
7 absolute numbers. But if all I knew about somebody is
8 that they're in this population and that they are under
9 30, I can multiple the population odds of 16 to one by
10 the information odds for that one isolated piece of
11 information of one to four, and the first line at the
12 bottom there, I find out that this person is four to one.
13 It's the 16 to one multiplied by one to four.
14 On the other hand, if I have somebody who is
15 over 50, you've got exactly the reverse situation.
16 Again, this is just a sample --an example. There 40
17 percent of the goods are over 50 and only 10 percent of
18 the bads. So the information odds there are the other
19 way around. It's four to one. Multiple that by the
20 population odds of 16 to one, and I can say people over
21 50 --my best guess, again, just based on this one piece
22 of information, age, are that they're 64 to one.
23 Well, now I've started spreading people out away

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 26.
26 Page 27 28
27
1 from that overall population odds of 16 to one. I've got
2 the younger borrowers over here riskier than the general
3 population, and the older borrowers a lot less risky than
4 the general population.
5 Of course, we don't look at just one factor. We
6 may throw in another factor, on the next slide, such as
7 own or rent, or residential status. And again, not
8 terribly surprising, the owners are better than average.
9 They're two to one. The renters are worse than average.
10 They're one to two.
11 And then on the next slide, I put those two
12 factors together and things started to get really
13 interesting, because now if I have an under 30 renter,
14 I'm taking again the population odds of 16 to one, the
15 information odds for under 30 of one to four, the
16 information odds for renter of one to two, and I multiple
17 that out, and now that under 30 renter is only two to one
18 odds. So if I lend money to three under 30 renters, I
19 can expect two of them to repay in a satisfactory manner
20 for every one that doesn't.
21 On the other hand, if I take my over 50
22 homeowner and multiple that out, now I've got odds of 128
23 to one. So I'm pulling different groups of that

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 27.
27 Page 28 29
28
1 population sort of further away from that center line by
2 adding the information value from these various factors.
3 Now, this example makes the assumption that all
4 of these factors are independent. Those of you who
5 remember the O. J. Simpson trial and the DNA evidence,
6 you'll remember there were some pretty staggering numbers
7 that the DNA experts cited of 9.5 billion to one odds
8 that these DNA markers could have belonged to anybody
9 other than O. J. Simpson.
10 Well, they were going through very much the same
11 process with a large number of individual DNA factors --
12 DNA characteristics, if you will. And based on the
13 assumption that they were independent, that's how you get
14 those kind of staggering numbers like 9.5 billion to one.
15 And if all of these factors in a credit decision were
16 also independent, you might get some pretty staggering
17 numbers.
18 That's not the way it works in credit. And in
19 fact, what we'll find is that there are correlations --
20 in some cases very high degrees of correlations --among
21 the different factors that we look at, so we've got to
22 correct for that. We don't assume that they're all
23 separate. We certainly wouldn't ever make a credit

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 28.
28 Page 29 30
29
1 decision using just age, or just own or rent, or just
2 those two factors, because in fact one of the things we
3 know --and it's pretty intuitive if you think about it
4 --is that as people get older, they're more likely to
5 become homeowners rather than renters.
6 You know, those of you who are over 30, you
7 know, remember back when you were in your 20's. Probably
8 most of us spent a few years as renters in our 20's, and
9 the chances of becoming homeowners get a lot higher as we
10 get older. And also, except for those who are in their
11 40's and still living with mom and dad, the time at
12 address is also going to go up as you get older.
13 And, you know, in a lot of cases, about the same
14 time that you get old enough to be legally able to borrow
15 anything, you're likely to leave home to go to college or
16 join the Navy and see the world or whatever, and so your
17 time at address takes a big hit. And then as you get
18 older, you're likely to be at the same address longer.
19 Homeowners, of course, are less likely to move than
20 renters, so there is a large correlation between time at
21 address and home ownership, as well as with age.
22 So these overlapping circles here, if you think
23 of the size of each circle independently as the

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 29.
29 Page 30 31
30
1 information value in that particular characteristic, the
2 overlap just suggests that because of these correlations,
3 while each of those factors taken individually may have a
4 fairly large information value, as you start adding these
5 correlated characteristics, you then have to correct for
6 that overlap.
7 And on the next slide, I throw in one more
8 factor. In this case, time on job. Well, of course,
9 time on job is likely to be correlated with those other
10 three factors. Again, unless your parents forced you
11 into child labor in the family business, if you're in
12 your 20's you're probably not going to have been at the
13 same job for very long. Your time at address and time on
14 job frequently go together, because if you get a new job,
15 in a lot of cases you've got to move, or maybe you have
16 an opportunity to move, and again, the home ownership.
17 So while time on job standing on its own has a
18 reasonable information value, what this represents simply
19 is that where you've already got those other factors --
20 kind of stability characteristics, age related
21 characteristics in the system --adding one more factor
22 of the same kind just brings very little additional
23 predictive value.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 30.
30 Page 31 32
31
1 And that's important for a number of reasons,
2 because typically in building a scoring system, we'll
3 start out with maybe as many as 50, and maybe as many as
4 a hundred, characteristics that independently have a
5 reasonable amount of information value. But because of
6 these correlations --these overlaps --as we actually
7 start building the system and we start getting a few
8 characteristics with a lot of information value in there,
9 a lot of those additional characteristics won't get into
10 the final system because of these overlaps.
11 Nevertheless, all of the information value there is still
12 captured by the characteristics that are in the system.
13 And so frequently somebody that's not sort of
14 familiar with the concepts of credit scoring comes along
15 and says, well, gee, this system doesn't have time on job
16 in it. I know that's predictive with credit performance.
17 There is something wrong with your credit scoring system.
18 No, actually the information value is in there, but it's
19 just picked up by those other factors that have
20 correlated with time on job. And so sometimes somebody
21 will come along and say, well, but you didn't consider
22 such and such. It's not in your scoring system. You
23 didn't consider such and such about me.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 31.
31 Page 32 33
32
1 Well, we may have considered that implicitly
2 because of all of these other factors that are in the
3 scoring system, and so for a number of reasons, they're
4 characteristics that people try to keep out of scoring
5 systems. One of them --something that, you know, sort
6 of might be a very obvious characteristic is income. It
7 sounds like a nice numerical characteristic. It's
8 probably got something to do with how people are going to
9 be able to pay their bills.
10 But it turns out that in practice income is
11 difficult to work with, because no matter how explicitly
12 the application asks the question, people insist on
13 answering it in different ways. You may ask for monthly
14 pre-tax income and you get semi-weekly, before tax and
15 after tax, and you get with and without child support and
16 alimony. And you get the rental property, even though
17 there is another box on the application for that. And it
18 actually turns out to be a very messy characteristic.
19 And so somebody says, well, you didn't ask for my income.
20 How could you possibly make an intelligent credit
21 decision. Well, it's because of these correlations.
22 If we go to the next slide, sort of the goal of
23 all of this, as I said, we're trying to separate --move

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 32.
32 Page 33 34
33
1 those people apart on the risk scale away from that
2 average that you know just because they walk in the door.
3 And what we really want to do in building this credit
4 scoring system is come up with a system that will
5 separate the eventual goods from the eventual bads by as
6 much as possible, so that --now there will always be
7 some overlap in the distributions.
8 Actually this doesn't quite do it justice,
9 because in reality the distribution of goods would go all
10 the way down here to the bottom of the scale, and the
11 distribution of bads would go all the way up to the top
12 of the scale. But what we're trying to do is move those
13 two distributions as far apart as we can, so that the
14 bulk of them in my nice bell shaped curve here --and of
15 course life isn't usually quite that perfectly drawn, but
16 it kind of approximates that.
17 What we're trying to do is move those two curves
18 as far apart as possible, so that we can set a cutoff
19 score that turns away as many of the bads as possible,
20 while turning away as few of the goods as possible. Now,
21 these are in percentages. Hopefully in any population
22 the goods would be a lot smaller in absolute numbers --
23 the bads would be a lot smaller than the goods in

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 33.
33 Page 34 35
34
1 absolute numbers. So these are in percentages, which is
2 why the curves are the same size.
3 But you can see at the top where the two curves
4 are a little further apart, you can set a cutoff that
5 cutoffs most of the bads and only a very small percentage
6 of the goods. Down at the bottom, the two distributions
7 are closer together, and so to turn away the same
8 percentage of bads, you've got to cut off twice as many
9 of the goods.
10 Now, that's the goal of the credit scoring
11 system. There is no absolute right or wrong, and we call
12 that measure divergence, just how far apart those two
13 distributions are. There is no right or wrong answer for
14 that. The goal is to get those two distributions as far
15 apart as possible in any given case, because then it lets
16 you make more efficient decisions at cutoff.
17 For example, for a college student population,
18 college students, you know, all look at lot alike in
19 terms of credit factors. Again, unless they're still
20 living at home, they're all renters. They've all been at
21 the address for a short time. If they've got a job at
22 all, they've been on the job for a very short time. They
23 have very little credit history, etc. So you might --if

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 34.
34 Page 35 36
35
1 you had a college student population, you simply couldn't
2 get the two curves as far apart as you could for a more
3 general population.
4 Next slide. Now, technically what credit
5 scoring systems do is to rank the order of consumers by
6 risk, because they don't consider things like economic
7 changes. If the economy goes south, people that scored
8 200 in this example that equaled odds of 20 to one, if
9 the economy takes a real downturn, those people that were
10 scoring 200 may find themselves with odds of 18 to one or
11 15 to one.
12 And so we don't purport to provide an absolute
13 odds quote, but, of course, what every credit grantor
14 wants to do is to turn that score into an odds quote.
15 And there are ways that they can do that, both from the
16 development statistics and also in terms of watching how
17 their own population performs.
18 So we get through all of this, and somebody is
19 probably still out there saying, oh, gee, but why do you
20 think scoring can do a better job of predicting credit
21 behavior of estimating risks than judgmental decisions.
22 It's real simple. It's better to count than to guess.
23 And fundamentally that's the difference between

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 35.
35 Page 36 37
36
1 how somebody using a credit scoring system is making
2 decisions. They're counting very carefully the prior
3 performance. They're taking a very hard look at the
4 prior performance of different kinds of people,
5 separating that into individual characteristics, and then
6 comparing that in a very disciplined way to what you know
7 about the next applicant through the door.
8 If somebody is making judgmental decisions,
9 they're trying to keep all of that in their head and, you
10 know, there is no way that you can juggle 10 or 12
11 different characteristics, each of which has three to six
12 different attributes. There is simply no way that you
13 can juggle that much information in your head, because if
14 you have 10 or 12 characteristics with three to six
15 attributes each, you have 10 to 20,000 possible
16 combinations in there. There is no way that somebody
17 judgmentally can keep straight 10 or 20,000 different
18 combinations of attributes.
19 And so to a large extent, they're guessing.
20 They're taking --making an informed guess, hopefully in
21 most cases. But really they're guessing at how today's
22 applicant compares to the applicants that they have seen
23 in the past and for whom they have actual performance.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 36.
36 Page 37 38
37
1 Okay. Real briefly I want to talk a little bit
2 about different types of credit scoring systems.
3 Historically most credit scoring systems were built for a
4 particular credit grantor, what we call custom
5 scorecards. The next category of borrowed scorecards is
6 really not terribly relevant these days, although in a
7 few cases it might be possible for one credit grantor to
8 borrow a system from a different credit grantor.
9 What is becoming very common, and again probably
10 what most people think of when they hear about a credit
11 scoring system, are credit scoring systems built from
12 data across multiple credit grantors. Now, of course,
13 the biggest example of that are the credit bureau
14 systems, that I'll talk about in more detail, and what we
15 call pooled system.
16 And then in a very few cases, somebody may come
17 along and want to do something kind of entirely new and
18 different. It doesn't happen in the United States,
19 again, very much these days. We do this more overseas.
20 Sometimes somebody will come along and want to do
21 something so new and different. They say well, we don't
22 have any data on this. We don't have any data on similar
23 portfolio. Fair, Isaac, you've been in this business for

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 37.
37 Page 38 39
38
1 40 years, what can you do for me.
2 And then in that case we can sit around the
3 water cooler and say, well, based on our 40 years of
4 experience in lots of different systems, here are some
5 guesses we can make about the kind of borrowers you're
6 going to see. We can build a judgmental --what's really
7 a judgmental scorecard, or what we call a launch
8 scorecard, just to get them going until they've got some
9 data.
10 Next one. The information used in a credit
11 scoring system can come from a variety of sources. Of
12 course, the most common are credit reports and credit
13 applications. A bank may have prior experience with a
14 particular customer. Sometimes demographic information
15 may be useful. For existing customers the issuer's
16 billing file has a lot of good information. And in
17 secured lending, the terms of the deal, the amount down
18 and things like that, can be very useful.
19 Next. Credit scoring can also be used to make a
20 variety of different types of decisions. Last year there
21 were about three billion pre-screened credit card
22 solicitations mailed in the United States. Probably 2.9
23 something billion of those were credit scored. Believe

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 38.
38 Page 39 40
39
1 it or not, not everybody in the world gets those pre-
2 screened solicitations, and typically one of the screens
3 they go through is credit scoring.
4 New applications, again, is probably what most
5 people think of in terms of credit scoring, making a
6 decision on an application for new credit. On the other
7 hand, if you pay attention if you own a credit card or
8 two, you'll notice that these days the issuers, for most
9 of you at least, try to do nice things once in a while.
10 They send you that letter that says because
11 you're such a wonderful customer, you don't have to make
12 the minimum payment this month. Or because you're such a
13 wonderful customer, we're going to increase your limit.
14 Or if you should happen to be standing at the cash
15 register with a purchase that's going to put you over the
16 limit, they want to know whether or not they're going to
17 authorize that transaction.
18 And finally when that piece of plastic expires,
19 they want to decide should we reissue it at all. If so,
20 should we reissue it for six months or for three years.
21 And so they're using decisions to manage those existing
22 --they're using credit scoring to manage those existing
23 accounts. They may also be using credit scoring to try

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 39.
39 Page 40 41
40
1 to decide whether or not to try to sell you another type
2 of product, based on what they know about your behavior
3 on the one you've got.
4 And finally, in certain kinds of lending,
5 especially mortgage lending, lenders frequently resell a
6 lot of the portfolio. The lender is using credit scoring
7 to decide which parts of the portfolio it wants to sell
8 versus which ones it wants to keep. And the buyer or
9 investor, on the other hand, may be using credit scoring
10 to decide which loans it is interested in buying.
11 There are also a number of different performance
12 definitions. Up until now I've sort of talked about
13 good-bad or really a scoring system designed to predict
14 any credit delinquency. We can --and there are systems
15 designed specifically to predict the risk of bankruptcy,
16 rather than other types of delinquency.
17 If somebody is making a mailing --a pre-screen
18 mailing --they want to know whether somebody is likely
19 to respond, as well as if they do respond and you issue
20 them a card, will they pay. There are a lot of people
21 who are absolutely great credit risks who are literally
22 not worth the price of a stamp for a pre-screen mailing,
23 because they are so unlikely to respond.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 40.
40 Page 41 42
41
1 And actually I'm kind of in that category. My
2 wife gets a lot more pre-screened mailings than I do,
3 because in fact we have sort of different patterns of
4 credit uses. And so somebody looking at her credit
5 bureau file is more likely to say this is somebody that
6 might be interested in another credit card. McCorkell
7 hasn't gotten another credit card in 15 years. We're not
8 going to waste a stamp on him.
9 Okay. Really just kind of an intro for the next
10 panel, scoring today in the mortgage industry, which is
11 probably as David and Jodie mentioned, is really how
12 scoring came to kind of the front of public
13 consciousness. Scoring is really used in a couple of
14 different ways. And, again, there is some terminology
15 here.
16 A lot of the mortgage industry refers --when
17 they say credit scores, they're referring to scores built
18 just on credit bureau data that are installed at the
19 credit bureau. When you order a credit report, a score
20 is generated. And when people in the mortgage industry
21 talk about credit scoring, they may mean what I would
22 probably refer to as bureau or credit bureau scores.
23 And most of those credit scores or credit bureau

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 41.
41 Page 42 43
42
1 scores used in mortgage lending, interestingly enough are
2 what I labelled here as generic scores, which are
3 designed to predict the risk of any credit delinquency.
4 There are also some mortgage-specific credit bureau
5 scores that were designed specifically to predict the
6 risk of mortgage delinquency. Actually what we found is
7 that the generic scores do a very good job of predicting
8 mortgage at risk, as well as other types of credit risks.
9 And then finally, what I've labelled
10 comprehensive mortgage scores, I think a lot of folks
11 just refer to as mortgage scores, where they are looking
12 not just at the credit bureau --the credit history piece
13 of information, but they're also looking at application
14 information and the deal terms --the amount down and
15 that kind of thing --in making a more comprehensive
16 decision. I think the next panel will get into that a
17 little more.
18 So what I would like to turn to next are
19 specifically the credit bureau scores. And here, first
20 of all just kind of broadly, are the types of information
21 that are considered by the credit bureau scores. Now,
22 these are scores that Fair, Isaac and others --there are
23 other scoring developers out there. I don't know why

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 42.
42 Page 43 44
43
1 anybody would use them, but some people do.
2 So there are other flavors of credit scores, and
3 occasionally somebody will use a FICO score to mean any
4 kind of credit score or any kind of credit bureau score,
5 even though FICO happens to be a trademark of ours. It's
6 a little like Kleenex. You don't always look at the
7 brand label when you walk into the store and buy a box of
8 Kleenex. You just want tissues to blow your nose. FICO
9 scores have sort of gotten the same kind of generic use
10 of the brand name.
11 These are scores that are running just on the
12 credit bureau data, and they're looking generally at five
13 types of information. And they're really listed here in
14 the order of importance. First is previous credit
15 performance. Has this person performed well, or have
16 they gone delinquent on other obligations. Not
17 surprisingly, that's the most predictive piece of
18 information we've got.
19 Secondly, now those for people where the answer
20 is yes, all of their prior credit they've performed in a
21 satisfactory way in the sense that they've met all of the
22 contractual obligations and they've never been late,
23 etc., that accounts for about 85 percent of the

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 43.
43 Page 44 45
44
1 population. And so there are only, again, rough numbers,
2 15 percent of the population that's got serious
3 delinquencies noted on their credit report at any given
4 time.
5 The other 85 percent of the population does not
6 all present the same degree of risk in terms of future
7 credit behavior. And so we've got to look at other types
8 of information as well to see how those folks are likely
9 to perform in the future.
10 And the next most predictive piece of
11 information is what's labelled, current level of
12 indebtedness. In one sense, you can sort of think of
13 that as how close to the edge are these folks. Is it
14 somebody --you know, somebody who has got five credit
15 cards, all of them they have borrowed right up to the max
16 and they're making minimum payments, are a lot riskier
17 than somebody that their balance --the limit ratio is
18 down in the 10 to 20 percent range.
19 The amount of time that credit has been in use.
20 And actually that line is a little bit mislabelled. It
21 probably ought to say sort of the amount of credit
22 history available, which is really a combination of the
23 time that credit has been in use and also the number of

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 44.
44 Page 45 46
45
1 trade lines that are on the file.
2 Now, trade lines can be on the file that aren't
3 active any more. In fact, again, those of us who have
4 been around for a while, sort of after we got into the
5 credit world, we probably have a lot of accounts that
6 we've paid off and closed. That good information, in
7 most cases, will stay on the file kind of indefinitely.
8 And so somebody that's got 20 or 30 years of
9 credit experience and any kind of reasonable number of
10 trade lines, maybe a lot of which have been paid off --
11 you know, student loans that have been paid off.
12 Mortgages that have been off. You know, your first
13 credit card may have been a gas card or a store card,
14 because that was the only place you could get a credit
15 card when you, you know, first got out of school.
16 So those have been paid off. All of those
17 historical pieces are likely to still be on the file, and
18 so we look at that. It's really the amount of credit
19 history available.
20 The next category we look at is the pursuit of
21 new credit. This is where the inquires come in and also
22 new account openings. Somebody that has a lot of
23 inquiries generated by the consumer --and I'll get into

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 45.
45 Page 46 47
46
1 that in a little bit --or has opened a lot of new
2 accounts recently, is more risky than somebody that
3 hasn't been out there actively pursuing a lot of new
4 credit.
5 If I could actually then skip down to I think
6 number --yeah, number 33. There are a few --I'm
7 skipping a few slides in the presentation, because we're
8 kind of running out of time here. Keep going. Keep
9 going. Bingo. Okay.
10 I mentioned inquiries. This has been a topic of
11 some controversy. Fair, Isaac about a year ago
12 redesigned our bureau systems to do what we call inquiry
13 de-duping. We know that in a lot of cases somebody
14 shopping for a mortgage or for an auto loan will wind up
15 with multiple inquiries from the same transaction. And
16 of course not many people buy more than one house in any
17 given month, or more than one car.
18 And so we've sort of been able to say, if we
19 count all of the auto loan inquiries and all of the
20 mortgage loan inquiries within any given 30 day period as
21 a single inquiry, because it's probably only one auto or
22 one house that they're shopping for, that that is really
23 a better representation of their credit behavior.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 46.
46 Page 47 48
47
1 And then also, again in mortgage lending
2 especially, people may wind up shopping around for some
3 period of time looking for the best rates, going to
4 different lenders or go through a mortgage broker who
5 shops at the different lenders. And so in fact what
6 we've said is, we don't count any inquiries in the 30
7 days before this particular score is computed.
8 So you can have a hundred inquiries in the month
9 before the score is computed and they won't count at all.
10 And you can have any number of mortgage or any number of
11 auto loans in any --I'm sorry. It's a 14 day period at
12 any time before the score is computer, and they'll only
13 count as one inquiry each.
14 Okay. If we could slip I think two slides.
15 It's just an example of the de-duping. One more. One of
16 the things we've done with the bureau scores is to
17 actually use multiple scorecards rather than a one size
18 fits all scorecard. And that will become relevant later
19 in the afternoon when we talk about what ought to be
20 disclosed.
21 And I think the question that Jodie posed is,
22 how can I raise my score. Because of the multiple
23 scorecard design, it's not possible to tell somebody, if

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 47.
47 Page 48 49
48
1 you do this, your score will go up by X, because in fact
2 the credit bureau information is a moving target.
3 There's always information coming into the credit bureau.
4 Essentially on a daily basis credit bureaus are getting
5 information from credit grantors.
6 Also, our systems consider the age of certain
7 types of information. A delinquency last week is a lot
8 worse than a delinquency five years ago. A new account
9 that was opened up last month is more significant than a
10 new account that was opened up three years ago in terms
11 of pursuit of new credit.
12 And so even if you don't do anything with your
13 credit. You don't use your credit card in a given period
14 of time and you don't make a payment. And of course if
15 you don't make a payment for too long, something will
16 happen, because the credit grantor will now start
17 reporting you as delinquent. But, you know, you don't
18 miss a payment, but you don't make a payment. Even if
19 you don't do anything with your credit history, just the
20 passage of time is going to affect some of those things.
21 And so I could sit there with somebody's credit
22 report and say well, if you close this account, or if you
23 paid off that account, here's what would happen to your

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 48.
48 Page 49 50
49
1 score if nothing else changed. But something else will
2 always change. And then one of the things that may
3 change, is that by doing something you may wind up on a
4 different scorecard, because the way that the system is
5 split, is first of all it looks for the presence of
6 serious delinquency information, either in the trade
7 lines or in the public record part of the credit file.
8 And so we have sort of the goods and the bads --
9 the previous goods and bads now separated using different
10 scorecards, because if we used a one size fits all
11 scoring system, anybody with any delinquency would get a
12 terrible score. But, in fact, not everybody with a prior
13 delinquency presents the same degree of risk, so we can
14 do a better job.
15 Now, the people with prior delinquencies as a
16 group are riskier than the people with no delinquencies
17 as a group. But we can do a better job by this split
18 scorecard design of finding the folks with prior
19 delinquencies that today pose a reasonable risk versus
20 the ones with prior delinquencies who are still just
21 terrible risks going forward.
22 And then depending --and then after we sort of
23 make the split based on the existence of delinquency,

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 49.
49 Page 50 51
50
1 then we also look at some other information: the
2 thickness of the file, in a sort of philosophical sense
3 how many trade lines, etc., there are, the age of the
4 file, how long has this person had established credit.
5 Again, somebody who is 50 years old and has 30
6 trade lines, most of which are now closed, that may be a
7 very typical pattern. Somebody who is 22 years old and
8 has 50 trade lines, and they've only been in the file for
9 24 months and they've got 50 trade lines already, that's
10 probably kind of a strange bird.
11 And so we're using different scorecards to
12 evaluate those different groups of people and the recency
13 of new trade lines opened. Again, somebody who is kind
14 of out there in a steady state mode who hasn't taken out
15 a lot of credit recently versus somebody who is kind of
16 out there shopping actively, we're using different
17 scorecards to evaluate them.
18 Now, the same score still means the same thing.
19 If you get a 700 on any of those score cards, it still
20 means the same degree of risk. But we get there by
21 looking at different information and maybe weighing
22 information differently, depending on which of those sub-
23 populations somebody is in.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 50.
50 Page 51 52
51
1 Okay. If you could again skip two. Just very
2 briefly, sort of the legal public policy issues, I guess
3 I'll save most of that for the panel this afternoon about
4 is it fair. I really just want to describe some of the
5 other slides that are in your packet.
6 We did two studies --and if you would go to the
7 next slide, please. We did two studies which are
8 labelled the LMI Study --low to moderate income. We
9 took data where we had --where we were developing custom
10 scorecards for specific credit grantors, and in many
11 cases where there were credit grantors who were going
12 from a judgmental decision process to a scoring
13 environment.
14 We looked at that data. We said sort of how do
15 the low to moderate income folks do when they're scored
16 versus how they were doing when the credit grantors were
17 making judgmental decisions. What is different about the
18 low to moderate income population versus the general
19 population, and sort of is scoring really an effective
20 way of making decisions on those populations.
21 A lot of people assume that for a lot of low to
22 moderate income borrowers, scoring just can't do a very
23 good job of evaluating credit risks, because they look

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 51.
51 Page 52 53
52
1 different than the rest of the population. Well, in
2 fact, what we found out was that not surprisingly the
3 lower income borrowers as a group scored lower. That's
4 why that yellow line is lower than the other lines in
5 this graph.
6 If you go to the next slide, what we also found
7 out is that at any given score, the risk of those low to
8 moderate income borrowers was in fact at least as great,
9 and usually in most cases just a little bit greater, than
10 the general population. So that if a score in this case
11 of 200 equaled the 20 to one for the general population,
12 for the low to moderate income population it probably
13 equalled 19 to one or 18 to one. In other words, they
14 were just a little more risky than the general population
15 in any score. So, in fact, the scores were doing a very
16 good job of predicting the risk for that population.
17 The other thing that we looked at --again, if
18 you would skip two. This is why they were different. If
19 you skip to the next one, there are two pages that look
20 very similar in your set, and unfortunately they are
21 mislabelled in the printed set. They're right on the
22 screen.
23 Slide number 42 should say "Maintain acceptance

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 52.
52 Page 53 54
53
1 rate; compare bad rates." And what we did here, is we
2 took the acceptance rate for the LMI population under the
3 prior judgmental decision process. We looked at the bad
4 rate that was coming up. And in fact, you see in some of
5 these cases --in fact, in a lot of these cases that bad
6 rate is terrible. It's horrible. The decisions that are
7 being made judgmentally on that LMI population are bad,
8 terrible decisions.
9 You know, the first couple of lines you've got
10 an acceptance rate of 26 percent and a bad rate of almost
11 11 percent. The next line you've got an acceptance rate
12 of 24 percent and a bad rate of 18 percent. Nobody is
13 going to live with an 11 or 18 percent bad rate in a bank
14 card portfolio. If you kept making judgmental decisions
15 there, what's going to happen is somebody is going to
16 come along and say, wait a minute. Our bad rate for low
17 income borrowers is totally unacceptable. We just can't
18 lend money to low income borrowers at all.
19 So what we did is say, well, okay. If you keep
20 the same acceptance rate and now you make those decisions
21 using scoring, what happens to the bad rate --and in
22 fact you'll see in every case the bad rate goes down, and
23 in many case very dramatically.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 53.
53 Page 54 55
54
1 The next slide, again, is mislabelled in the
2 printed set. It's okay up here. It should say "Maintain
3 bad rate; compare acceptance rates." Now, again,
4 maintain the bad rate for most of these is probably not a
5 realistic scenario. But using scoring, what you see is
6 if you were willing to live with that bad rate --and
7 actually the third and the fourth ones, the bad rates may
8 be acceptable, 2.3 and 4.3 percent.
9 But the acceptance rates judgmentally were very
10 low for those folks. If you evaluated that same
11 population using credit scoring and you're willing to
12 live with that two and four percent bad rate, you could
13 accept in the one case --in bank card number four you
14 could accept four times as many applicants as you were
15 accepting judgmentally. And that's again very consistent
16 across that.
17 The next study that we did was using credit
18 bureau information. And there, rather than looking at
19 the low to moderate income population, we looked at
20 consumers in zip codes with a high concentration of
21 minorities, blacks and hispanics. And I could get into a
22 lot detail about why we chose the various categories we
23 did. But basically we looked at zip codes with those

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 54.
54 Page 55 56
55
1 high minority concentrations. One of the things we found
2 was that those zip codes represented a little bit less of
3 the credit bureau database than they did of the entire
4 population, although certainly they weren't totally
5 unrepresented, as I think a lot of people assume.
6 And indeed there are folks in that population
7 that simply don't use mainstream credit. They are never
8 going to have a credit report. But they're never going
9 to go to a lender that would pull a credit report in the
10 first place, and secondly the degree of under
11 representation isn't so great that it would really affect
12 the ultimate scorecard design.
13 And we see there --if you would go to the next
14 slide --that in fact that zip code definition winds up
15 looking a lot like the low to moderate income definition,
16 because those zip codes have approximately --the average
17 household income in those zip codes --and this was a few
18 years ago now --is about two thirds of the average --
19 or, I'm sorry --of the household income of the remainder
20 of the population. So not surprisingly in economic
21 terms, those high minority zip code populations look a
22 lot like the low to moderate income populations, and so
23 not surprisingly we came up with very much the same

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 55.
55 Page 56 57
56
1 results.
2 If you go to the next slide, again you'll see
3 the score distribution is a little bit lower, which again
4 means that at any given cutoff score, you're going to
5 accept fewer of those folks. But if you go to the next
6 slide, you'll see again exactly the same pattern. In
7 this case the red line is the minority population. The
8 dotted yellow line is the rest of the population. At any
9 given score those folks represent a slightly higher
10 degree of risk.
11 So again, the scorecard is doing an effective
12 job of rank ordering risk, and to the extent there is any
13 discrepancy in the risk estimation, it actually favors
14 those high minority areas of borrowers just as it did the
15 low income borrowers. Again, for the credit grantors
16 listening, that difference is very slight. It's not
17 worth getting excited about. But to the extent there was
18 any difference, it was consistently in favor of the low
19 to moderate income and the minority zip code borrowers.
20 Again, we looked at how those two populations
21 differed, and finally we asked ourselves, well, could we
22 build --next slide. Could we build a scorecard
23 specifically for that minority population that worked

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 56.
56 Page 57 58
57
1 more effectively than the scoring systems that are being
2 used now.
3 Well --and next slide, please. What we found
4 out was that if we used a one size fits all credit bureau
5 score, in fact then we could build a more effective
6 minority zip code scoring system. Now, I suspect that
7 the folks from the Fed and some of the other regulatory
8 agencies might want to have a conversation with us if we
9 did that. But this was a research exercise. We found
10 that it would be possible.
11 But what we also found was that the existing
12 Fair, Isaac credit bureau score using the multiple
13 scorecard approach, where you're looking at different
14 sub-populations separately, where you're looking
15 separately at the folks with prior delinquency, and
16 you're looking separately at the folks that don't have a
17 lot of established credit, that that multiple scoring
18 system that we were using was in fact more effective in
19 predicting risk in that minority zip code population than
20 a scorecard --a single scorecard developed specifically
21 for that population.
22 I've run a little bit over my time. I thank
23 everybody for your indulgence, and I'll see you this

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 57.
57 Page 58 59
58
1 afternoon. Thank you.
2 MR. MEDINE: Thank you, Pete, for an excellent
3 overview of a very complex issue.
4 (Applause.)
5 MR. MEDINE: Now we're going to take a minute
6 and get our next panel set up here, and then we'll resume
7 in about two minutes.
8 (Brief recess.)
9 MR. MEDINE: We just heard from Pete McCorkell
10 on the issue of credit scoring generally. This next
11 panel is going to focus on the use of credit scoring
12 specifically in the mortgage industry. Obviously home
13 purchases are a consumer's largest purchase. And so how
14 credit scores interrelate with that decision is of
15 critical importance to consumers. And the presentations
16 will also include information about mortgage scoring and
17 the automated underwriting process.
18 We're going to start from my extreme right, so
19 I'll first introduce Pamela Johnson from Fannie Mae. Pam
20 is Vice President of Single Family Mortgage Business for
21 Fannie Mae. She is responsible for new product
22 enhancements to Fannie Mae's automated underwriting tool,
23 the Desktop Underwriter, and is responsible for credit

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 58.
58 Page 59 60
59
1 research and data acquisition for Fannie Mae.
2 We're thrilled to have her here this morning.
3 Thank you, Pam.
4 MS. JOHNSON: Thank you. Can everybody hear me
5 okay? All right. Are we ready? Let me start first of
6 all by stepping back for a minute, because I think I want
7 to clarify and make sure everybody understands where we
8 are here.
9 Peter did a fabulous job walking through how
10 does one develop a scorecard. But I want to be clear
11 that there are many different types of scorecards out
12 there. There are many different --they're used in a
13 variety of industries, not just in the mortgage industry.
14 In fact, the scorecards have been around a lot longer,
15 and used for a longer period of time, than they have in
16 the mortgage industry.
17 The consumer industry --for example, credit
18 cards, etc. --have been using scorecards, I don't know,
19 how long, Peter, 20, 30 or 40 years. But in the mortgage
20 industry it's only been recent that we've adopted the use
21 of credit scoring as a tool in underwriting.
22 And I want to be clear about which scorecard we
23 recommend, which is the use of the credit bureau

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 59.
59 Page 60 61
60
1 scorecards. So Peter spoke a little bit about this in
2 his presentation. And I wanted to point out that the
3 characteristics in the credit bureau scorecard are those
4 listed on page 24 of Peter's presentation, and not the
5 ones that he talks about early on, which could relate to
6 other scoring models.
7 So in case you missed it, he listed the five
8 characteristics that are most prominent in a credit
9 bureau scorecard. And those were previous credit
10 performance, current level of indebtedness, amount of
11 time credit has been in use, pursuit of new credit and
12 types of credit available. So just as a way of
13 clarification.
14 First of all, let me say I'm delighted to be
15 here. I thank David and team for inviting me. I think
16 this is a critically important issue, and I think this is
17 a terrific forum.
18 Credit scores and automated underwriting are a
19 very important tool in the mortgage industry. At Fannie
20 Mae we spent a considerable amount of time researching
21 and analyzing credit scores. And long before we
22 determined that they would be useful in mortgage
23 underwriting, we went through a lot of work to answer

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 60.
60 Page 61 62
61
1 some very important questions. Those questions are: do
2 they work, are they predictive of risk as it relates to
3 mortgage lending, are they fair, do they work for all
4 borrowers in the same way, and when used correctly, will
5 they expand home ownership opportunities.
6 We answered yes to all of those questions after
7 a lot of research, and I want to take you through some of
8 our research and some of the analysis we did to get
9 there.
10 The first slide, please. So today I'm going to
11 talk to you about credit scoring, mortgage scoring --
12 because they are different --and automated underwriting.
13 I'll also discuss how scores should be used to originate
14 and underwrite mortgage loans.
15 Next slide, please. As Peter indicated, credit
16 scores are empirically derived in a statistical method of
17 assessing risks. A credit score --and specifically,
18 again, we're talking about a FICO credit bureau score --
19 is based solely on the information in the borrower's
20 credit report. The evaluation of the information in the
21 borrower's credit report, meaning how the borrower
22 manages credit, has always been an important factor in
23 mortgage underwriting. Credit scores when used correctly

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 61.
61 Page 62 63
62
1 are based on information in the credit report that has
2 been proven to be predictive of credit risk and loan
3 performance.
4 Next slide, please. Mortgage scores, on the
5 other hand, are slightly different. A mortgage score is
6 also based on information that has proven to be
7 predictive of credit risk. A mortgage score is based on
8 information about the borrower obtained from the loan
9 application and the credit report, as well as information
10 about property value.
11 Mortgage scoring supports a comprehensive
12 analysis of both the borrower's ability to repay a loan
13 and the borrower's management of credit. A mortgage
14 score helps an underwriter perform a more consistent,
15 objective and accurate evaluation of this information
16 than traditional underwriting does.
17 Next slide, please. Fannie Mae recommends that
18 lenders use credit scores, specifically the FICO credit
19 bureau scores, in their manual underwriting process. In
20 addition, we have developed a more comprehensive
21 customized mortgage scoring model for use in Desktop
22 Underwriter, our automated underwriting system.
23 Next slide, please. Okay. So why should we use

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 62.
62 Page 63 64
63
1 credit scoring in mortgage lending.
2 Next slide, please. Credit scores are in fact
3 predictive of credit risks for all loans and all
4 borrowers. A borrower with a credit score below 620 is
5 in the neighborhood of two and a half and sometimes three
6 times more likely to default on a mortgage loan than
7 someone with a credit score between 660 and 699.
8 Next slide, please. Okay. This slide
9 illustrates this point. And if you'll look at the green
10 box in the middle, which indicates a score --a loan with
11 a score range of 660 to 699, and you compare that to a
12 loan with a score range of less than 620, you can see the
13 dramatic difference in the likelihood of default
14 performance.
15 Next slide, please. Similarly, this slide shows
16 how default rates on mortgage loans increase as credit
17 scores fall. Specifically, this chart shows that for
18 every 60 points in a decrease --a 60 point decrease in a
19 credit score, there risk of default doubles. If you'll
20 look at the blue line, the blue line is in fact a
21 geometric curve that shows a doubling of risk every 60
22 points, and the red lines are the actual default rates by
23 a FICO score. And you can see how closely correlated

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 63.
63 Page 64 65
64
1 they are.
2 Next slide, please. We have learned from the
3 analysis of loans delivered to Fannie Mae that a
4 borrower's management of credit, as measured by a credit
5 score, had little correlation with the borrower's income.
6 In other words, it's just as likely for a low income home
7 buyer to have a high credit score as a high income home
8 buyer.
9 Specifically, 39 percent of the borrowers whose
10 mortgage loans were delivered to Fannie Mae in 1998 and
11 who had incomes of 80 percent or less of the area median,
12 also had scores over 740. The percentage of borrowers
13 with incomes over 180 percent of area median who had
14 credit scores over 740 was 37 percent.
15 I think most of you probably know that a 740
16 credit score represents an excellent credit risk and an
17 excellent credit history, and can be used as a
18 compensating factor for other high risk factors in a
19 mortgage loan, such as a low down payment.
20 Next slide, please. Okay. The predictiveness
21 of the credit scores allows us to adopt a different and
22 more dynamic view of the risk of borrowers who make low
23 down payment loans --make low down payments. This is a

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 64.
64 Page 65 66
65
1 view of the relationship between FICO credit scores and
2 loan to value ratios. Traditionally, loan to value
3 ratios have served as an important indicator of mortgage
4 default risks with high LTV loans representing a higher
5 degree of default risk than low LTV loans.
6 This slide shows that when you combine credit
7 scores with loan to value ratio, there is an even greater
8 opportunity to expand home ownership. A borrower who
9 puts less than 10 percent down but has an excellent
10 credit history, as here defined as a credit score above
11 740, is less likely to default than a borrower who puts
12 30 to 40 percent down with a poor credit history.
13 These numbers in this chart --these are
14 relative numbers. We use as the number one what we call
15 an average loan, which is indicated --you can see the
16 one in red in the middle, which here is a loan with a 76
17 to 80 LTV and a credit score between 660 and 699. All of
18 the other numbers on the chart are related to that one,
19 meaning they perform either better than or worse than
20 that average loan.
21 Next slide, please. Credit scores are an
22 objective measure of credit risk based on information
23 from credit reports that have proven to be statistically

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 65.
65 Page 66 67
66
1 predictive of credit risk and loan performance. Fannie
2 Mae has verified that the credit scores are predictive of
3 mortgage loan performance for all loans and for all
4 groups of borrowers regardless of demographic
5 characteristics such as race or income.
6 Next slide, please. Okay. This slide
7 illustrates this last point. No matter what the race of
8 the borrower, the lower the credit score, the greater the
9 likelihood of default. And we've done this slide over
10 numerous different time periods, and the results continue
11 to be consistently the same.
12 Next slide, please. Let me tell you what Fannie
13 Mae's policy is on the use of credit scores in mortgage
14 underwriting. We advise lenders that credit scores
15 should be used as a careful --as part of a careful,
16 comprehensive analysis of each mortgage application.
17 They should not be used in isolation or as the sole
18 factor in making an underwriting decision.
19 This is the policy that we have communicated
20 with our customers --our lender customers --and we've
21 made that point clear in our lender letters that we've
22 issued on credit scoring and how to use it.
23 Next slide, please. Fannie Mae purchases loans

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 66.
66 Page 67 68
67
1 with credit scores below 620. We do not prevent lenders
2 from delivering these loans to us, nor do we make them
3 re-purchase these loans simply because the credit score
4 is below 620. Our mortgage scoring model development
5 data sets, the data sets for the scoring model, include
6 loans with scores below 620, making it possible for our
7 current mortgage scoring models that we use in Desktop
8 Underwriter to accommodate these loans and accurately
9 assess their risk.
10 Desktop Underwriter will underwrite and approve
11 loans with scores below 620 and does so. In our letters
12 to lenders on credit scoring, we have provided specific
13 examples of situations in which it would be acceptable to
14 sell us loans with credit scores below 620.
15 Next slide, please. Okay. Over the past
16 several years, Fannie Mae has developed and made
17 available to mortgage lenders our automated underwriting
18 system called Desktop Underwriter.
19 Next slide, please. Desktop Underwriter has
20 been designed as a decision support tool for lenders. It
21 helps reduce origination time and costs through
22 streamlined processes and opportunities to re-engineer
23 work flow.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 67.
67 Page 68 69
68
1 For example, Desktop Underwriter facilities
2 point of sale underwriting decisions. Desktop
3 Underwriter also helps lenders make more consistent,
4 objective and accurate underwriting decisions. However,
5 just as with credit scores, the use of Desktop
6 Underwriter is meant to compliment the underwriter's
7 judgment. It does not replace the underwriter's
8 judgment.
9 Desktop Underwriter performs a comprehensive
10 risk assessment of each loan application. This risk
11 assessment is much more comprehensive than a credit
12 score. It is based on the mortgage scoring model,
13 combined with the rules taken from Fannie Mae's selling
14 guide to determine loan and borrower eligibility.
15 Next slide, please. Desktop Underwriter, or DU
16 as we call it, issues one of three recommendations for
17 most loans. An approved recommendation, which means that
18 the loan meets Fannie Mae's credit underwriting
19 guidelines and does not need further review by an
20 underwriter. A referral recommendation, which means that
21 the loan may need Fannie Mae's credit underwriting
22 guidelines, but it needs to be reviewed by an underwriter
23 for that determine. Referral with caution

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 68.
68 Page 69 70
69
1 recommendation, which means that the loan does not appear
2 to meet Fannie Mae's credit underwriting guidelines, and
3 it must be reviewed by a human underwriter.
4 And it is in fact the referral and referral with
5 caution loans that we look to our lender partners and
6 underwriters to work with. That's the message we send.
7 Don't deny it; work with the loan. Those are the loans
8 that need extra attention.
9 Next slide, please. Desktop Underwriter does
10 not approve or decline loans. Underwriters make those
11 decisions. Fannie Mae has not used any prohibited
12 factors, such as race, age, gender, color, religion,
13 national origin or marital status, as characteristics or
14 risk factors in the mortgage scoring models or risk
15 assessment rules in Desktop Underwriter.
16 Next slide, please. As part of our ongoing
17 effort to validate that our scorecard works in our
18 mortgage scoring model, we consistently look at
19 performance. Now, this slide --this chart shows the
20 percentage of loans that become seriously delinquent, and
21 that means they've missed three payments, by the age of
22 the loan and the DU recommendation.
23 Now, the reason why we look at serious

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 69.
69 Page 70 71
70
1 delinquencies as opposed to defaults is that we don't
2 have enough time or history yet with our loans that have
3 been delivered after having been underwritten in Desktop
4 Underwriter. You need several years of history with the
5 loans --seasoning with the loans --to really track
6 default performance.
7 So we're tracking delinquency performance right
8 now, and a loan that misses three payments is considered
9 a high risk loan for foreclosure and default. After 12
10 months the percentage of the serious delinquencies is
11 more than six times greater for the loans designed refer
12 with caution than for the loans recommended for approval
13 in Desktop Underwriter.
14 Next slide, please. Desktop Underwriter
15 evaluates the information in a mortgage application based
16 on factors that have proven to be statistically
17 predictive of credit risk and loan performance. DU
18 evaluates the information in each mortgage application
19 the same way each time, consistently, objectively and
20 accurately, and without regard to the demographic
21 characteristics of the borrower.
22 Next slide, please. Desktop Underwriter does
23 not use --and I mentioned this before --any of the

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 70.
70 Page 71 72
71
1 characteristics such as race, age, color, religion,
2 national origin or marital status as factors in the
3 scoring model or the risk assessment rules in DU. In
4 addition, where the borrower lives and where the borrower
5 is purchasing a home are factors that are also not used
6 in the mortgage scoring model.
7 Fannie Mae has verified that DU's
8 recommendations are predictive of mortgage loan
9 performance for all borrowers --for all groups of
10 borrowers --for all loans and regardless of
11 characteristics such as race and income.
12 DU provides opportunity for expansion or
13 broadening of the market by looking at an array of
14 factors that can compensate for other high risk factors
15 in the loan. When used properly, and when the
16 information in the credit report is accurate --and I
17 want to emphasis that --credit scores and mortgage
18 scores and automated underwriting support consistent and
19 objective underwriting decisions.
20 And when you look back at where we've come, for
21 years we've used human underwriting and manual
22 underwriting. And it is my opinion that manual
23 underwriting creates a greater opportunity for

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 71.
71 Page 72 73
72
1 discrimination because of the subjectivity associated
2 with manual underwriting. One of the benefits of using
3 credit scoring and using mortgage scoring is it does
4 provide an objective analysis of the likelihood of
5 default or the credit risk of that particular borrower or
6 loan.
7 And that concludes my remarks.
8 MR. MEDINE: Thank you very, very much. Let me
9 just --if I could just ask a quick question --and Pete
10 eluded to this earlier --about the comparative benefits
11 of judgmental versus credit scoring.
12 You mentioned one benefit in terms of the
13 possibility of injecting discrimination into the process?
14 MS. JOHNSON: Um-hum.
15 MR. MEDINE: Have you also analyzed whether
16 certain people are more likely to qualify for, say, a
17 loan under a credit scoring system than under a
18 judgmental system? In other words, does it benefit more
19 borrowers to be scored than to have been subject to a
20 judgmental system?
21 MS. JOHNSON: I think if used correctly. Again,
22 the way we recommend that scores and mortgage scoring is
23 used, is to really take the loans that are sort of the

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 72.
72 Page 73 74
73
1 easier ones to do and put those aside, and really work
2 with the loans that are the harder loans. And what we've
3 found is that when you combine credit scoring and other
4 factors to get a mortgage score, that you actually have
5 increased your opportunities for making loans to more
6 borrowers.
7 Specifically to your question, David, when we
8 looked at the slides around the combination of LTV and
9 the FICO score, and that, you know, a 10 percent down
10 mortgage was in fact performing better than some that --
11 where someone may have put down 20 or 30 or 40 percent,
12 when you overlay credit scores on top, I think that gives
13 you an example of the opportunity that exists to expand
14 the numbers of loans one can make.
15 Because no longer is a low down payment loan by
16 itself the highest risk alone in the bucket. And in
17 fact, it may be --there may be plenty of opportunity to
18 expand into, you know, a greater population of low down
19 payment loans by using credit scores as a compensating
20 factor.
21 MR. MEDINE: Thank you very much. And Peter is
22 nodding, so why don't we turn next to Peter Mahoney, who
23 is the Associate General Counsel at Freddie Mac. Peter

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 73.
73 Page 74 75
74
1 is responsible for legislative and regulatory matters,
2 with a particular emphasis on housing policy and fair
3 lending issues at Freddie Mac.
4 He's a frequent speaker on housing, fair lending
5 and community reinvestment issues, and has published
6 several articles on community reinvestment and housing
7 issues.
8 MR. MAHONEY: You can stop reading now.
9 MR. MEDINE: Thank you for joining us.
10 MR. MAHONEY: That's the whole version there.
11 It goes on. I've got a home page that's got the whole
12 thing, rotating head and everything.
13 (Laughter.)
14 MR. MAHONEY: I'm going to stand up, because I
15 tend to have a lot of nervous tics, and so this helps me
16 get rid of it. I ran for an hour this morning. But this
17 is a pretty big group, and it's also a very impressive
18 group. I recognize a lot of the experts in the industry.
19 A lot of regulators. So I'll stand up and move around a
20 little bit and give the cameraman something to do here.
21 It doesn't surprise me. Pam and I didn't really
22 get a chance to talk about our presentations too much,
23 because we are fieriest competitors and this is the FTC,

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 74.
74 Page 75 76
75
1 and I didn't run the risk of an antitrust violation. But
2 I doesn't surprise me that Fannie and Freddie are saying
3 some of the same things. So my presentation may trod on
4 a little bit of the same ground, so I'll try to go
5 quickly over those issues.
6 I thought since I knew I was going second and
7 that Pam is so confident and well spoken and does such a
8 great job, that she would hit some of these highlights,
9 so that I could maybe concentrate on some of the touchier
10 questions regarding the use of credit scoring in mortgage
11 lending, and particularly focus on what we think of as a
12 lot of the questions we've heard since 1995 when we first
13 started talking about this, a lot of the questions that
14 we've heard from the community.
15 We do do a lot of talking and then listening
16 about credit scoring. And we've heard kind of the same
17 questions over and over, so I thought I would put
18 together for you, as soon as Mr. Clark is ready --you
19 can roll the next one. This is a logo that I have to put
20 on everything I do.
21 The myths associated with credit scoring --and
22 I didn't bring copies of this handout, but David Jamison,
23 who is a legal intern with us this summer from Howard Law

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 75.
75 Page 76 77
76
1 School, is the designated person to whom you give your
2 business card or your web address or whatever, and we'll
3 get a copy of this to you. In addition, we had some
4 materials out front. There is a report that we did in
5 1996 on automated underwriting, and issues regarding
6 fairness and consumer issues. I think those copies are
7 all gone, so we can get a copy of that to you if folks
8 have not already seen that.
9 Let's go through these myths as quickly as
10 possible. You can go ahead and roll through. These are
11 the ten myths, but I think I'll just get to them as we
12 go, because I know we're constrained for time. And I do
13 give you an instruction to cut me off, because I am
14 Irish.
15 Go to the next slide. Myth. Only one type of
16 credit score is used in mortgage lending. I think, as
17 Pam aptly pointed out, and I think Pete pointed out the
18 same thing, there are a lot of different types of credit
19 scoring models that are used in the mortgage market.
20 Pete talked at length about FICO scores. The Fair, Isaac
21 score uses only what is in one of the three national
22 credit repositories. So it is only going to deal with
23 credit performance of a consumer. It doesn't include

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 76.
76 Page 77 78
77
1 kind of personal type of information, etc. It also
2 doesn't include attributes that are uniquely associated
3 with a mortgage loan.
4 Most mortgage lenders want to know how much
5 you're going to put down for the house and what the
6 collateral value of the house is. They're going to want
7 to know what type of loan you're looking for. Is it a
8 one year ARM? That's riskier. Is it a 30 year fixed
9 rate mortgage loan? That's less risky. They're looking
10 for characteristics beyond your credit and payment
11 record.
12 FICO only looks at credit and the payment
13 record. We have our own team of crack scientists at
14 Freddie Mac. We have just --our Chairman is an
15 economist. He buys economists. We have so many
16 economists, you wouldn't believe it. We have a huge
17 group of folks who do nothing but estimate scorecards all
18 day at Freddie Mac.
19 And we have built our own custom mortgage
20 scoring model, and we have deployed it in a system called
21 Loan Prospector. It was the first automated underwriting
22 system made available broadly for mortgage lenders, and
23 it is a service that uses a custom mortgage scoring model

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 77.
77 Page 78 79
78
1 that is based not only on the information that is in a
2 credit record kept at the repositories. We go and get
3 those and bring them in. But it's also based on a host
4 of other variables that I'll discuss with you in a few
5 moments.
6 Okay. We strongly endorse the use of automated
7 underwriting. We think that's kind of the A way to
8 underwrite mortgage loans today. However, for lenders
9 who are still using manual underwriting systems --in
10 other words, they're still going down their kind of
11 written checklist and qualifying people kind of the old
12 fashion way today --we strongly recommend that they also
13 pull a generic FICO score from each of the three
14 repositories.
15 And then we tell them to throw out the high one,
16 throw out the low one, and use that middle one as a good
17 benchmark for getting you into assessing someone's credit
18 worthiness. Like Pam said, we tell people don't use it
19 exclusively. But we think it can be used as one tool in
20 the manual underwriting process.
21 Okay. Second, credit scores make lending
22 decisions. And the question David asked I was interested
23 in, because the fact of the matter is, credit scores are

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 78.
78 Page 79 80
79
1 a tool. It's like going to the gas station and saying,
2 you know, is my battery fluid low. And they put
3 something in and a little dial shuts up and says
4 zzzzzzzz. You know, that's the score of your battery.
5 That's not going to fix anything. That's not
6 going to make a lending decision. That's not going to do
7 anything for anybody. It is simply a ranking tool. And
8 once the result of the credit score is made available to
9 someone who has money in their pocket, then the lender
10 makes a decision on the basis of the information they've
11 learned from the credit score and from other things.
12 How lenders --and I include Freddie Mac and
13 Fannie Mae as a lender. How we apply credit scores, what
14 decisions we make in using credit scores, and how we
15 deploy them, that is where you get all of the questions
16 about what is the effective impact of credit scores. Is
17 it constricting credit, or is it going to expand markets.
18 And I'll talk a little bit more about that later, but I
19 did want to drive that point home.
20 Now, what it does do, is it permits you when
21 you're making decisions to do them consistently and
22 objectively without bringing into account other factors
23 that may raise significant issues about intentional bad

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 79.
79 Page 80 81
80
1 will toward a particular applicant.
2 Credit scoring models are a black box and we
3 don't know how they work. Well, I would beg to differ
4 with that, and I think this is probably one of the most
5 controversial myths. A lot of people are going, well, I
6 still don't know how they work, okay. So credit scoring
7 models rely on common and publicly available variables.
8 Pete went through, and Pam told you, page 24 of
9 Pete's slide presentation, that shows you the type of
10 variables that are used. And by variable I mean just an
11 input in the FICO generic bureau score. Okay. Now, you
12 can get better than that. You can contact each of the
13 three repositories, and they will give you a list of the
14 feedback messages --and there's a long laundry list of
15 them --that each of the three repositories use when they
16 send a credit score. That gives you a very, very good
17 idea of the type of information that the Fair, Isaac
18 model is looking at when it creates one of these generic
19 credit scores.
20 In addition, Freddie Mac has published
21 information about the variables that we rely on. We rely
22 on all those credit inputs. We also use loan devalue,
23 which is how much money is someone putting down on a

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 80.
80 Page 81 82
81
1 house. That's a very, very powerful indication. If
2 someone has one percent down on a house versus 40 percent
3 down on a house, they're going to have a very different
4 risk profile. And we use a lot of other kind of mortgage
5 related factors.
6 Let me point out one thing, though. With all
7 due respect to Pete, on page seven and eight of his
8 presentation, he listed a bunch of potential application
9 type variables. And I just want to state categorically
10 for the record that Freddie Mac does not use in any of
11 its custom scoring models time at present address, time
12 at present job and residential status. We don't use age.
13 We don't use income. We do not use, do you own or rent,
14 years at address, occupation or any of that stuff.
15 MS. JOHNSON: Ditto for Fannie Mae.
16 MR. MAHONEY: Okay. I just want everyone to get
17 that down. We're aware of some factors. A first time
18 home buyer is a predictive factor. We don't use it,
19 because our statutory purpose --one of them --is to
20 expand home ownership opportunities for people who want
21 to go from renting to owning. So we don't use that.
22 And even our written underwriting guidelines
23 have been modified heavily in the last ten years to

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 81.
81 Page 82 83
82
1 eliminate some of those old customs and practices of
2 mortgage underwriting, and to say, don't look at how long
3 someone has been on a job. Look at what their income
4 projection is for the next couple of years. That's kind
5 of the way we look at things.
6 Okay. We can go to the next one now. This
7 gives you a little snapshot of how the credit score model
8 that we deploy in Loan Prospector works and how it's kind
9 of related to standard, traditional underwriting criteria
10 for mortgages. We use collateral, down payment. We use
11 credit reputation, history of repayment, current account
12 balances, inquiries and new accounts. The same kind of
13 thing that a FICO score measures.
14 We use a debt to income ratio. We look at how
15 much cash reserves you have when your down payment is
16 done and you've closed your deal. Those factors are then
17 cooked in a big mathematical equation, with guys and gals
18 in white lab coats. And I don't understand them and I
19 don't get any of it.
20 But I do understand what the variables are, and
21 I understand that what they produce then is a set of
22 inputs that creates a very comprehensive statistical
23 profile of how --what your odds are of defaulting on

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 82.
82 Page 83 84
83
1 your mortgage in the future, based on the information
2 going in compared with this very wide range of other
3 folks. And it's as simple as that.
4 Now, credit scores are not based upon a
5 representative population sample. I think this is an
6 issue that's largely going to be going away, because we
7 have confirmed that generic credit scores --the stuff
8 that's in FICO scores that we've recommended --are based
9 upon a representative sample based on some of the stuff
10 that Pete showed you, which is that there is a 6.7
11 percent on minority area representation versus an 8.3
12 percent of the overall population. We think that's very
13 good, given how many folks have credit records, etc.
14 For our own custom mortgage scoring models, we
15 have ensured that there is a representative sample used.
16 We look at demographic characteristics. We look at
17 geographic characteristics to try to make sure that when
18 we're kind of creating the sample from which we build the
19 scorecard that those kind of factors are used.
20 But the proof of the pudding is in the tasting.
21 And the representative --we believe representativeness
22 is best demonstrated by the fact that, as I'll show you,
23 credit scoring models that we use are predictive for

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 83.
83 Page 84 85
84
1 those neighborhoods and for those folks, and therefore we
2 think that that is a very strong indication of fairness
3 in the fact that it's taken into account on enough
4 different folks' behavior.
5 Credit scoring has not been independently
6 validated. That is not an accurate statement. The
7 Federal Reserve Board --actually Bob Avery is here, and
8 you can ask him about this. They obtained some credit
9 scores. They tested it against low income populations,
10 as well as a broad sample of populations. Their findings
11 were published in the July 1996 Federal Reserve Bulletin.
12 I'm still looking for my autographed copy.
13 Recently --very recently HUD economists have
14 been working for a couple of years with Fair, Isaac on
15 estimating their own scorecard based on FHA borrowers.
16 FHA borrowers, as you know, tend to be much lower
17 income and to have a higher concentration of minority
18 households, particularly African American and Hispanic
19 households.
20 They recently reported at a Fannie Mae research
21 round table that mortgage scorecards are successful in
22 predicting risks for FHA borrowers. There is also
23 evidence that mortgage scorecards are predictive within

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 84.
84 Page 85 86
85
1 subgroups such as low income and minority borrowers. And
2 that's an article that was put together by three
3 economists at HUD, so that's a very strong, independent
4 corroboration.
5 And then we've published our own stuff, which I
6 told you about, this automated underwriting report which
7 came out in 1996. And I'm about to show you some
8 additional information.
9 Credit scoring is unfair to low and moderate
10 income borrowers. Now, Pam showed you her statistics,
11 and mine don't look much different. We both buy millions
12 of loans, so they probably wouldn't look different.
13 Actually, FICO scores and Loan Prospector scores that we
14 have do not vary much by income. And I think the reason
15 why is that income is a very lousy proxy for wealth.
16 There are people who live in Beverly Hills who are low
17 and moderate income, because they live in a census track
18 that happens to be very high income.
19 So income is a --as Pete said is a notoriously
20 difficult thing to measure. So while we take some
21 comfort from the fact that we think across --you know,
22 we try to drill down into that and confirm that our
23 scoring models work for truly low income neighborhoods

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 85.
85 Page 86 87
86
1 where the median income is very low, and they do. But I
2 guess what I'm suggesting is they're flat across the
3 board, simply because income is a notoriously difficult
4 thing to track.
5 And now I'm going to show you a picture. Yeah.
6 This is based on our 1994 mortgage purchases. This is a
7 Loan Prospector scoring model, okay, and what we did is,
8 we just cut this thing up into tenths. And we're showing
9 you, these are the high risk guys over here, and these
10 are the low risk guys over here as identified by our
11 scoring model, and these are the actual default rates.
12 Okay. This is from 1994, and we measured this earlier in
13 1999. Over a million loans. And these are low income,
14 middle income and high income folks. And I can define
15 that for you if you want me to, but trust me.
16 And what you're seeing is a very, very similar
17 distribution, okay. So the folks that are identified by
18 our model as low risk, actually are barely defaulting.
19 Those folks in the last three buckets particularly, very
20 high risk and very high default rates. Anyone in the
21 mortgage business knows if you're approaching an eight or
22 10 percent default rate, you're in a lot of trouble.
23 Next one. The next myth is that credit scoring

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 86.
86 Page 87 88
87
1 is unfair to minority borrowers. I would like to put on
2 the table once again and emphasize something I think we
3 looked past. Over the last ten to 15 years the industry,
4 with a lot of help from public policy and housing
5 advocates, has made an enormous amount of progress in the
6 way we treat a broad cross section of borrowers who want
7 to purchase a home. We have made enormous strides, and
8 there is a lot of statistical evidence to back that up.
9 And I think that one of the things we loose
10 sight of when we talk about credit scoring and the
11 progress we've made, and how rapid technological change
12 is, is that we're saying, you know, oh gosh, you know, I
13 have problems with credit scoring. Well, the solution
14 would be to go back to the way we used to be doing it,
15 where you had a single loan officer kind of deciding
16 whether a person got a loan. I don't think anyone wants
17 that system.
18 Now, as to the statistical results, again,
19 credit scores very accurately predict borrower default,
20 regardless of race or ethnicity, and in our view, again,
21 ensure a level playing field.
22 We can go to the next one. Again, this shows
23 you the demographic results. This is 1994 borrowers.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 87.
87 Page 88 89
88
1 This is just a sample. But I can tell you that I've seen
2 a lot of samples and this is quite representative of the
3 African American households, Hispanic households and non-
4 minority households.
5 Now, you do see a slightly higher default rate
6 as you get really all across here. And again, what
7 accounts for that, I'm not completely sure. I believe
8 it's a wealth factor. If I have an income of $150,000
9 and you have an income of $150,000, but I'm the first
10 person in my family that ever went to college and I'm
11 supporting my grandmother, my mother and I'm helping out
12 over family members, if I lose my job, I'm going to have
13 a higher likelihood of defaulting.
14 If I make 150 grand and I lose --I have the
15 same income as this other person. I lose my job, but I
16 can call up my mother and say, hey, can you give me
17 $5,000, I need to make some mortgage payments and pay my
18 credit cards, those are the kinds of things that we
19 really can't measure by income statistics. But I think
20 they do have a profound effect on some of these results,
21 and it is a generational issue that, you know, God
22 willing will obviate over time.
23 Go to the next one. Here is a myth that I --

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 88.
88 Page 89 90
89
1 you know, I hope people recognize. Credit scores are
2 always predictive for every single person. That is not
3 true. And so I know we're going to hear from some folks
4 who believe that credit scores were not fair and that
5 there were perverse results created by use of credit
6 scoring.
7 I am not up here to attempt to deny that there
8 are folks who could pay off a home mortgage loan or some
9 other trade account, and the credit score may indicate
10 that they are higher risk than they believe themselves to
11 be or that some other piece of evidence they have would
12 indicate. There are inaccurate credit records out there,
13 and the FCRA and other statutes are there to correct
14 that.
15 But even beyond that, a credit score is simply
16 going to say your odds versus a bunch of other people
17 across the spectrum or that you're a higher risk. And
18 that is not a moral condemnation of those folks. It is
19 what we believe to be a very accurate picture. And so
20 again, all that we're saying is that credit scoring
21 improves the odds that we're going to get it right, okay,
22 in predicting mortgage performance. Moreover, we believe
23 that over time our models will get better. We're

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 89.
89 Page 90 91
90
1 collecting more and better information on folks, and this
2 first generation of models should not be thought of as
3 fixed in stone.
4 People without credit scores cannot obtain a
5 mortgage. As Pam said, that's just not true. We buy a
6 lot of below 620 FICO scores. That number comes from the
7 industry letters that Fannie and Freddie published in
8 1995, and some folks had believed that lenders were going
9 to turn that into kind of a bright line below which you
10 are sent into the darkness to wail and gnash your teeth.
11 But we buy a lot of particular loans that have
12 gone through automated underwriting. We will balance off
13 the risks, and if someone has a dinged up credit record
14 indicated by a credit score, that can be offset by
15 compensating factors. In addition, if there are
16 particular life circumstances when we refer the loan back
17 that a borrower can work through with the lender, we will
18 buy those loans.
19 We have a lot of underwriting guidelines
20 designed to deal with extenuating circumstances and
21 difficulties in life history that mean that a credit
22 score is not indicative of the actual future mortgage
23 performance. And we take those seriously, and we buy a

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 90.
90 Page 91 92
91
1 lot of mortgage loans on that basis.
2 Now, we also work with a lot of people. I know
3 Marcia Griffin is here. I don't see her, but I know
4 she's here from Home Free USA. We work with a lot of
5 credit counseling agencies and housing counseling
6 agencies. And again, people can reestablish credit or
7 establish credit for the first time within a couple of
8 years, and that to me is one of the best roads to go,
9 which is get a good track record and then get the
10 responsibility of home ownership under your belt.
11 Next slide, please. Okay. And finally, credit
12 scoring reduces the availability of low cost mortgage
13 credit. I can tell you that there is an article right
14 here that David actually pulled off the Lexis/ Nexis for
15 me, which estimated that the efficiencies of automated
16 underwriting systems have cut costs for originators which
17 have been passed onto consumers in the form of lower fees
18 and points at closing.
19 And this guy from Nomura Securities, whom I've
20 never met, said that closing points and fees in July of
21 last year on a 30 year mortgage averaged 1.1 percent
22 versus 1.6 percent in 1993. Okay. So points have gone
23 down. Closing points have gone down. It takes four

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 91.
91 Page 92 93
92
1 minutes to approve a mortgage loan application now,
2 whereas in 1993 we were still dealing with a couple of
3 weeks for many applicants, okay.
4 We're talking about efficiencies. We're a macro
5 economic deal, Freddie and Fannie are, and what we're
6 trying to do is make the system more efficient. And
7 scoring allows us to bring technological efficiencies to
8 bear. And we can predict risks, which means we can creep
9 down on that risk spectrum. We can offer our low cost
10 financing to more people because we understand their
11 risks.
12 And we're doing that right now. We're buying
13 loans right now today that are under automated
14 underwriting that are scored that a year ago or two years
15 ago would have been called A minus. And those people
16 have a lower mortgage interest rate because of our
17 ability to tap low cost funds.
18 So we believe that over time credit scoring will
19 dramatically --dramatically --reduce costs, and that
20 most of the folks who have been shut out of the prime
21 market and into the sub-prime market are a
22 disproportionately higher share of those are low income
23 and minority folks. So our fervent belief is that this

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 92.
92 Page 93 94
93
1 is an excellent and promising development for mortgage
2 markets and that scoring properly understood is a great
3 tool to achieve our objective of expanding markets.
4 And that's all I got. Thank you.
5 MR. MEDINE: Thank you very much for your high
6 energy presentation. We next turn to Carroll Justice
7 from FT Mortgage. Carroll began his mortgage banking
8 career in 1984 and is currently Executive Vice President
9 at FT Mortgage Companies, where since 1992 he has served
10 as head of Secondary Marketing with responsibility for
11 interest rate management, private investor applications,
12 and private investor and agency relations.
13 Carroll?
14 MR. JUSTICE: Thanks, David. Good morning. I
15 want to thank you all for the opportunity to allow FT to
16 talk about an issue that has become very critical and
17 very beneficial to our industry.
18 FT Mortgage Companies is a Dallas-based
19 residential A quality mortgage lender. In 1998 we
20 originated roughly 24 billion dollars in mortgage loans.
21 We have 167 offices in 31 states. Our originations are
22 55 percent retail and 45 percent wholesale. We service
23 loans in all 50 states. We have over 450,000 loans in

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 93.
93 Page 94 95
94
1 our servicing portfolio.
2 Sixty percent of our loans are sold to Fannie
3 Mae and Freddie Mac. Twenty percent of our loans are FHA
4 and VA loans and are securitized in Ginny Mae Mortgage
5 Back Securities. And the remaining loans are sold to
6 private investors. As a typical mortgage banker, 100
7 percent of our loans are sold into the secondary market.
8 We portfolio no loans that we originate.
9 We began using credit scores, and in our case
10 FICO scores that we talked about this morning, with the
11 release of Freddie Mac's Bulletin 96.6 in September of
12 1996. With this Bulletin Freddie Mac strongly --and
13 I'll emphasize that --strongly recommended the use of
14 credit scores on all loans we sold to them. Fannie Mae
15 quickly followed suit, and now all of our significant
16 investors require some form of credit score usage. That
17 is, with the exception of FHA and VA.
18 Now, to facilitate loans moving from one program
19 to another, even though all of our investors don't
20 require them, we require a FICO score as part of the
21 credit report on the loans we originate. Now, if a
22 credit score is in dispute or not available, we use other
23 credit information and totally disregard the credit

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 94.
94 Page 95 96
95
1 score.
2 Disputed credit scores represent basically one
3 percent of our originations. Loans with no credit scores
4 are roughly one to two percent. So in effect we're able
5 to generate a credit score of somewhere between 97 and 98
6 percent of the loans that we originate.
7 Now, besides meeting investor requirements, we
8 also use credit scores because we found them to be a very
9 good predictor of credit quality. To validate this, we
10 took a look at our 1998 originations. And what we did,
11 is we looked at those loans that went 90 days into
12 default within the first 12 months or origination. Now,
13 we picked 90 days, because it's a very significant level
14 of default in a very short period of time.
15 In looking at these loans, we went into our
16 servicing system and tried to make sure --or I shouldn't
17 say tried. We made sure there was no noise coming out of
18 our servicing system, such as loan not being set up
19 properly or the borrower not receiving coupons. Any
20 reason or any time the loan went into suspense in our
21 servicing system, we threw that loan out.
22 Looking at that population and that criteria, we
23 found that loans with FICO scores below 585 were ten

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 95.
95 Page 96 97
96
1 times greater to go into a 90 day default rate in the
2 first 12 months than loans with FICO's of 660 or higher.
3 Loans with FICO scores between 585 and 620 had a seven
4 times greater probability of going into default.
5 Now, when you consider the cost of default
6 services for a mortgage banker like us, it can be as high
7 as 40 percent of the cost of servicing. The average
8 servicing cost probably runs between 60 and 70 dollars a
9 loan per year. We, with over 450,000 loans in our
10 servicing portfolio, think that credit quality is a very
11 important item in our originations.
12 A third reason for using credit scores is that
13 they are a critical component of automated underwriting
14 systems. With the introduction of Fannie Mae's Desktop
15 Underwriter and Freddie Mac's Loan Prospector, mortgage
16 lenders had new tools that allowed them to use the power
17 of the computer to analyze a very complex credit and
18 collateral decision that was required to make a sound
19 credit decision.
20 Before automated underwriting, manual processes
21 followed rules based models. If the loan didn't fit the
22 rules, it could get rejected or at least covered up with
23 paper to cover the underwriter's decision. This not only

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 96.
96 Page 97 98
97
1 increased the time and the cost to originate a loan, it
2 also increased the probability the loan could get
3 rejected.
4 After using automating underwriting systems for
5 two years, it's our experience that AU --or automated
6 underwriting --makes it easier to get a loan. It not
7 only reduces processing and underwriting times. It
8 reduces documentation requirements and also makes a
9 better credit decision than manual processing and
10 underwriting. You know, the additional documentation
11 that's required of traditional underwriting by itself can
12 cause borrowers to become discouraged and perhaps even
13 give up on the process.
14 In 1989 we pioneered a limited documentation --
15 alternative documentation is probably a better term --an
16 alternative documentation quick closing system. We did
17 this with the approval of Fannie Mae and Freddie Mac.
18 What we had found that caused us to do this, is that our
19 industry typically made the loan process and underwriting
20 requirements a very onerous process for most borrowers.
21 And we decided that whatever we could do to reduce the
22 documentation and the time required would improve the
23 likelihood that more borrowers would get loans.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 97.
97 Page 98 99
98
1 Roughly --or currently, I should say.
2 Currently roughly 80 percent of our loans --of our
3 conventional loans --go through DU or LP. We do not
4 have a proprietary system of our own. We are currently
5 using LP on our FHA and VA loans, and we were one of the
6 pilots for the Fannie Mae DU system for FHA loans.
7 Now, in addition to the agencies, many of our
8 nonconforming investors accept DU or LP for loans sold to
9 them, or they at least have their own proprietary system.
10 What we've seen over the last two years is a widespread
11 expansion of automated underwriting throughout the
12 mortgage finance industry. And a critical component of
13 that automated underwriting has been the use of credit
14 scores.
15 I want to stop here just a moment and kind of
16 explain how we use credit scores, or how we use DU and
17 LP. We gather the information. If a loan is an accept
18 on either of the systems that we use, the loan is
19 automatically approved. If a loan is a caution or a
20 reject, it then goes through our manual underwriting
21 process. What the automated underwriting system does for
22 us on the caution loans, is that it allows us to focus on
23 a specific area and makes the process much easier to

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 98.
98 Page 99 100
99
1 decide whether or not we have a good credit or we have a
2 bad credit.
3 Now, I understand that many of you question the
4 use of automated underwriting and credit scores because
5 of their potential for disparate treatment. Our
6 experience does not support your concern. We found a
7 five to ten percent increase in the number of high risk
8 loans that get approved with the use of automated
9 underwriting. We believe this is due to the system's
10 ability to examine all data available and quickly compare
11 compensating factors. A human underwriter using
12 traditional rules based systems can't make the same
13 complex decision on what has historically been a high
14 risk loan.
15 At FT the credit score is usually the --the
16 credit score is usually first available at the time of
17 underwriting when the credit report is ordered.
18 Technology constraints currently restrict us from viewing
19 it at the point of sale. In spite of this, we are,
20 however, moving towards a point of sale automated
21 origination system. In the future many of our customers
22 will know their rate, price and underwriting decision
23 when they meet with a loan officer.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 99.
99 Page 100 101
100
1 Now, the important point to remember here is not
2 how technology is going to be used. However, it is that
3 we don't use credit scores to pre-qualify, but rather as
4 a part of the total credit picture. Credit scores are
5 used at the underwriting decision point, not to pre-
6 screen applications. No borrower at FT is denied a loan
7 because of a credit score. As I said earlier, if there
8 is any question about the validity of a score, or no
9 score exists, we manually underwrite the loan using
10 traditional credit information.
11 Now, to answer how these disputed loans or no
12 credit score loans perform, we've also looked at that.
13 And what we have found is that borrowers with no credit
14 score have this 90 day, early payment default rate,
15 slightly worse than our average EPD rate. Borrowers with
16 incorrect or disputed scores have default rates roughly
17 equal to those default rates on borrowers with FICO
18 scores less than 620, or roughly seven times our 660
19 rate.
20 Now, in addition to not using disputed credit
21 scores, we don't have cutoff levels on FICO scores for
22 our loan products unless the investor requires it. For
23 conventional or FHA loans with FICO's under 585 that are

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 100.
100 Page 101 102
101
1 manually underwritten, we do require two levels of
2 underwriting approval. That's not only to make sure
3 we're getting a good loan, but it's to make sure that we
4 don't have disparate treatment against somebody who has a
5 low FICO score. If DU or LP approves the loan, we accept
6 that approval, even though it may have a low FICO score,
7 without any further underwriting.
8 After using credit scores for roughly three
9 years, we have made the following conclusions about
10 credit scores and automated underwriting. Credit scores
11 have helped us identify superior quality loans, and
12 therefore will help control future servicing and
13 origination costs. This will help us lower our overall
14 costs to produce and therefore our pricing to customers.
15 As Peter pointed out, we feel that the lower
16 costs --lower operating costs associated with credit
17 scores and automated underwriting will be passed on
18 directly to the consumer. For those of you who are very
19 familiar with mortgage bankers, you'll know there are no
20 subsidies that we don't like. If we have an opportunity
21 to pass it on to the borrower, we're going to pass it on
22 to the borrower primarily for competitive reasons.
23 We have found that a higher percentage of our

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 101.
101 Page 102 103
102
1 high risk applicants are being approved because of a
2 combination of credit scores and automated underwriting,
3 because this provides a superior credit, analysis and
4 underwriting decision to manual underwriting.
5 We now look at credit differently than we used
6 to. We used to just consider payment histories. Now,
7 the amount, frequency and recent use of credit are just
8 as important, if not more so. We now have a credit
9 measure that helps us equate credit quality with the cost
10 of default. Credit abusers will be penalized by driving
11 up the cost of housing for themselves. Credit scores
12 will help us develop an automated risk-based pricing
13 model that will allow more credit to become available to
14 high risk borrowers.
15 And finally, the use of credit scores in
16 conjunction with automated underwriting will help lenders
17 become more efficient in making loans, investors more
18 efficient in providing funds for the housing industry,
19 and improve the ability of borrowers to get housing
20 credit.
21 Thanks.
22 MR. MEDINE: Thank you. Peggy?
23 (Applause.)

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 102.
102 Page 103 104
103
1 MR. MEDINE: Peggy Twohig is the Assistant
2 Director for Financial Practices, has a question for
3 Carroll.
4 MS. TWOHIG: Carroll, I just wanted to make sure
5 I understood correctly one thing you said. Did you say
6 you do not use credit scores at the pre-qualification
7 stage at all?
8 MR. JUSTICE: No.
9 MS. TWOHIG: You do not?
10 MR. JUSTICE: No. What we've found, Peggy, is
11 that with a lack of technology to be able to deliver a
12 credit score into the origination process, we found that
13 we're better off using it as part of the total credit
14 report and using it in conjunction with automated
15 underwriting.
16 MS. TWOHIG: So your pre-qualification decisions
17 are judgmental only?
18 MR. JUSTICE: Yes. For the most part they are.
19 MS. TWOHIG: Okay. And do you know if that's
20 typical among lenders?
21 MR. JUSTICE: I'm not sure if it is or not.
22 MS. TWOHIG: Okay. Thanks.
23 MR. MEDINE: Okay. I want to thank Pam, Peter

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 103.
103 Page 104 105
104
1 and Carroll for excellent presentations. We're going to
2 take a 15 minute break and resume with consumers'
3 experiences with credit scoring. Thanks again.
4 (Brief recess.)
5 MS. TWOHIG: As David mentioned, my name is
6 Peggy Twohig. I'm one of the Assistant Directors with
7 the Division of Financial Practices with the Federal
8 Trade Commission.
9 Good morning everyone. I have the pleasure of
10 introducing some panelists who are going to talk about
11 some of these issues from the consumer perspective. We
12 had excellent presentations this morning about what the
13 perspective is on credit scoring from the industry's
14 view, what credit scoring is and how it's used in
15 mortgage lending. And jumping ahead, also we heard some
16 perspectives that address some of the issues we're going
17 to talk about this afternoon. And that was very helpful
18 in laying the ground work.
19 So now we would like to move to credit scoring
20 from the viewpoint of consumers and those that help
21 consumers with mortgage lending. Is anything wrong with
22 this picture? What are the questions? What are the
23 concerns? What are the issues that they see in credit

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 104.
104 Page 105 106
105
1 scoring?
2 And we're going to let our three panelists
3 speak, and then we would like to open it up to you all.
4 We are running a little bit late, but I think we can
5 still accommodate questions and issues raised by the
6 audience. Our purpose here in this session this morning
7 is to raise the issues --ask the questions and get all
8 the issues and questions out on the table. And then this
9 afternoon we'll try to make sure that our panelists
10 address those questions.
11 Our first panelist on this panel is Simi Batra,
12 who is representing himself as a consumer who first
13 encountered credit scoring several months ago when he
14 began thinking about purchasing a home. Simi?
15 MR. BATRA: Good morning. First of all, I want
16 to put a disclaimer here. I speak for myself, not for my
17 employer on this matter. I don't expressly represent his
18 views.
19 I am planning on buying a house in the District.
20 And about a year ago --almost a year ago I ordered my
21 credit reports. Everyone tells me, clean up these things
22 way ahead of time. It takes a while. Look for
23 inaccuracies. And I did that. And it wasn't such a

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 105.
105 Page 106 107
106
1 tough endeavor cleaning up the credit reports.
2 But now after this process --or most of the way
3 through this process, I understand that --I was under a
4 crazy impression that a mortgage banker was going to open
5 up my credit reports, go through my payment history here,
6 and ask me questions about my payment history. And I
7 guess that's not the case.
8 This is going to boil down to a number. I'm not
9 sure who boils down this number, if it's the credit
10 bureau or if it's the mortgage broker himself or herself.
11 But my credit history is going to be boiled down to a
12 number. I have no idea where that number comes from or
13 what it means. It will be rated on a scale. I'm not
14 privy to this number. I'm not privy to the scale. I'm
15 not privy to what creates this number.
16 So I meet with a mortgage broker in two days. I
17 guess I'll find out my number then and see if I was
18 successful over the last year in doing the right things
19 and everything I was supposed to do.
20 Along those lines, I have a few points to make
21 with regard to credit scoring. Some are pretty specific
22 and some are more general. Generally proponents of
23 credit scoring have two opposing things they're talking

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 106.
106 Page 107 108
107
1 about. One, they say we don't need to know the specific
2 factors of what goes into a credit score. If we knew
3 these factors, we might be able to manipulate our credit
4 scores and that would be bad.
5 But on the other hand, if a credit score is
6 indicative of our future repayment, how can it be so
7 easily manipulated? It doesn't really make sense to me.
8 If this is supposed to be a seven year history of your
9 credit repayment, and based on these solid factors and
10 this is how you're going to pay your mortgage in the
11 future, why does it really matter if I know what goes
12 into a credit score then?
13 I shouldn't be able to change these things.
14 It's predictive. It's a matter of a credit card balance
15 or a ratio. I mean, it shouldn't be that arbitrary. So
16 it's kind of --it's an opposing thing you get with the
17 credit score folks.
18 And there are many --one thing I'm kind of
19 concerned about is, I called some mortgage brokers on the
20 phone. I say what's your latest balance --or what's
21 your latest interest rate, and they'll say --they'll ask
22 me the question, what's your credit score? A lot of the
23 mortgage brokers themselves have no idea that we are not

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 107.
107 Page 108 109
108
1 privy to credit scores. Not only are we not privy to it,
2 the law says they don't have to give us our credit scores
3 specifically.
4 And a lot of the mortgage brokers themselves
5 assume that these credit scores are coming from the
6 credit bureaus. I don't think a lot of them realize that
7 they themselves --their company --has ordered these
8 credit scores. They're paying for them. And they kind
9 of think it comes with the consumer and we know what our
10 credit scores are. I mean, there's 40 or 50 credit
11 scoring models out there. I have no idea and I don't
12 know what my credit scores are.
13 So one thing, I think part of this needs to be
14 education of the mortgage brokers and those folks. Let
15 them know that consumers do not know this information.
16 We are not privy to this information.
17 Okay. What goes into a credit score? That's
18 something that I'm concerned about. I'm not really
19 worried about my payment history, or where I've lived, or
20 how long I've been at my job. Things that I can't
21 change. More things like credit card balances, the
22 balance of your debt to your available credit, the number
23 of credit cards you have open.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 108.
108 Page 109 110
109
1 Over the last year in cleaning up my credit
2 reports, what I've tried to do is consolidate my credit
3 card debts. You know, that department store thing I
4 opened ten years ago to get 10 percent off, I forgot I
5 even had it. You know, I ripped it up right away, but I
6 forgot it was still open. So I just wanted to close that
7 out.
8 But in doing so, the economy being so good, I've
9 gotten a hundred credit card offers in the mail with
10 really low interest rates. So what I did, was I --like
11 a lot of people --consolidated all my debt unto one
12 credit card. And as I'm going through this process of
13 consolidating my debt, I'm assuming day by day as I cut
14 up a credit card, or close out one or open one to move
15 it, my credit score is changing, because the ratio of my
16 debt to my available credit changes and the number of
17 open credit cards I have changes.
18 At some point in there, and over the last year,
19 I probably had an ideal credit --an ideal balance of
20 available debt --debt to available income or credit and
21 the perfect amount of open credit cards. I don't know
22 what that perfect balance is. And somehow this is
23 supposed to be predictive of my future repaying my

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 109.
109 Page 110 111
110
1 mortgage.
2 Well, if it's so predictive --it just seems
3 very arbitrary. If I applied for a mortgage half way
4 through this process, maybe I would have had the perfect
5 ratio then. I don't know. Maybe I have it now. I don't
6 know. And the point is, it's not predictive. It's
7 arbitrary. It's entirely arbitrary. It's not really
8 based on any scientific factor.
9 It's something you can change so easily. I can
10 call my credit card company tonight and ask for a greater
11 balance. You know, greater available credit. And maybe
12 they'll give it to me. And if I do, that will change my
13 credit score. I'm not sure that's really predictive of
14 anything, or it's a matter of me asking for the right
15 amount of --it's just a matter of luck.
16 Now, I have a couple real specific points with
17 regard to credit scoring. I mean really specific, and
18 it's part of --I don't know what goes into the charge
19 cards. For example, American Express. These things
20 don't have balance limits on them. Now, what are they
21 using for a balance. I mean, these things are getting
22 plugged into my credit score. I have no doubt, including
23 my payment history, which is easier. It's late payments

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 110.
110 Page 111 112
111
1 or it's not late payments.
2 But in terms of your high credit limit, your
3 available credit and things like that, I'm under the
4 impression from looking at my three credit reports that
5 charge cards --they use the highest amount you've ever
6 charged in a single month and it's essentially viewed as
7 your credit limit. That's kind of what I'm thinking
8 about. Whereas your current balance is what you have on
9 there now, and the difference between the two is your
10 available credit.
11 Now, if that's the case, that's good to know.
12 But maybe I should have known that six months ago, and I
13 would have charged a whole bunch on a charge card, paid
14 it off, and this month charged about 20 percent of that
15 amount, so I have that perfect little ratio that I don't
16 know. What if I were getting married next week and my
17 charge card --this was the month I maxed it out. Maybe
18 I should have done it a month ahead of time and then paid
19 it off. I don't know. It's arbitrary and I just wish I
20 knew that.
21 And one of the credit bureaus does not report
22 American Express correctly. They do not --what they do
23 is, every month they change --they essentially put your

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 111.
111 Page 112 113
112
1 credit --your debt is equal to the highest you've ever
2 owed. And that's --so it almost looks like you have 100
3 percent debt ratio in there. And that's problematic if
4 that is going into my credit score. I don't know,
5 because I don't know what goes into a credit score.
6 And the last specific point I want to make about
7 credit scoring is inquiries. Now, this concept is so
8 ridiculous to me. And this is something that I
9 encountered and it was very problematic. I opened up my
10 credit report and there were like 20 inquiries for auto
11 insurance. I mean, I must have called every 800 number I
12 saw a commercial for on TV to get the best auto insurance
13 rate, not knowing --no one told me, by the way, this
14 phone call will lower your credit score. This phone call
15 will lower your credit score. I wish I had known that.
16 Inquiries are treated as applications for
17 credit, whether or not you were just inquiring in a
18 dictionary sense as to just asking a question, what is
19 the rate you will give me, as opposed to actually I
20 applied for credit. Well, I'm not applying for credit.
21 I just want to know what's the rate you're going to give
22 me.
23 And I wasn't aware that a lot of these auto

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 112.
112 Page 113 114
113
1 insurance companies will look at your credit reports to
2 do that, and they will then show up on the inquiry
3 section in your credit report as an application for
4 credit. And I encountered that problem and had many of
5 those removed, which is another problem with this. It's
6 not easy to remove inquiries from your credit reports.
7 It's very hard to remove.
8 Credit bureaus will not investigate them as they
9 do a normal late payment. Therefore, the obligation is
10 up to the consumer to do that. And to try to call an 800
11 number that really doesn't exist, it's just --they don't
12 exist. I mean, people do not work behind these numbers.
13 And it's unbelievable the amount of time you have to
14 spend to get an incorrect inquiry off your credit report.
15 It takes forever. It's a multi month process.
16 And once you do finally get that letter from
17 that auto insurance saying we did not mean to look at Mr.
18 Batra's credit file, the credit reporting agencies will
19 say, well, we will take up to 90 days to then correct
20 that information. So you spent, you know, six months on
21 this process, and then they'll tell you they have another
22 90 days to correct that information.
23 I find that a bit problematic, and especially

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 113.
113 Page 114 115
114
1 since I don't know how that's going to be affecting my
2 credit score. And 90 days is a long time just to correct
3 information that's already been presented to them.
4 What's really ridiculous about the insurance --
5 the auto insurance thing is, you apply --you do a price
6 quote for insurance, that goes on your credit. But once
7 you get the insurance policy, that's not on your credit
8 report.
9 (Laughter.)
10 MR. BATRA: So it's strange. Asking Geico how
11 much they're going to charge you for insurance shows up
12 as an inquiry, but your insurance policy doesn't. So it
13 just completely makes absolutely no sense to me. And I
14 think that's a glaring example that can be fixed rather
15 easily. All they have to do is code an inquiry versus an
16 application for credit. Did you actually apply for
17 credit, or did you just ask them what the price would be.
18 And there is a big difference between the two.
19 And that would be a simple coding in any formula. It's
20 just these formulas seem to be so rigid and they're
21 probably very much --they're very long. Not a simple
22 thing to do and very expensive, but something that needs
23 to be done. And that's a very simple thing I think they

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 114.
114 Page 115 116
115
1 should do, and rather quickly.
2 But these are a few of the things I've come
3 across in cleaning up my reports. And the concept of
4 credit scoring is foreign to me and to everyone I've
5 talked to at work, my friends, my roommates and my
6 family. No one has heard of it. Everyone is like, I pay
7 my bills on time. I have no credit card debt. Well,
8 there's one more step there that you don't know about.
9 And maybe blissful ignorance is good if you have
10 a good credit score. You never even knew that there was
11 such a thing. I have friends who have bought houses.
12 They never knew about a credit score. That's all right.
13 I mean, everything worked out okay for them. Maybe it
14 was fine. But it just should be a little more brought
15 forth to the public. Let them know this thing exists.
16 It's a tool. It's only a tool, but some
17 mortgage brokers are using it as more than a tool. It's
18 kind of a lazy way out to say here's a number, yes or no.
19 Does it meet our criteria, yes or no. And it should be
20 just a tool and I think it's being abused a bit. And I
21 think some education in the lending sector and the
22 consumer sector is really necessary.
23 Thank you.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 115.
115 Page 116 117
116
1 MS. TWOHIG: Thank you, Simi. I think
2 everybody should know that Simi took time away from his
3 very demanding professional job to come and talk about
4 his experiences, so he did not have the benefit of
5 hearing what Pete McCorkell said this morning about
6 inquiries and how they're counted and not counted. But I
7 think actually this is almost better, because he can say
8 what his impression is from he just knew as a general
9 member of the public, or didn't know or understand, or
10 does not understand about credit scoring.
11 Moving to our next panelist, Caryl Iseman is the
12 owner of Action Mortgage Group in San Diego, California.
13 Caryl has 26 years of experience in real estate and
14 mortgage financing, concentrating on finance in the past
15 nine years. In the process of helping borrowers attain a
16 home loan or refinance their current loans, and in
17 counselling them and helping them with their credit
18 problems, Caryl has encountered several issues with
19 credit scoring that she will identify for us today.
20 MS. ISEMAN: I, too, have to stand up. I spent
21 a lot of time sitting on the phone flying from San Diego.
22 So forgive me. I tend to move around a little bit, too.
23 And I want to say thank you very much to the Federal

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 116.
116 Page 117 118
117
1 Trade Commission for putting on this panel. Because what
2 my hope is, is that at the end of this day with the
3 presentations that are being made, that we will walk out
4 with the benefit of knowing (1) what credit scoring is,
5 but (2) with potentially some resolutions.
6 The system is broke. I am sorry to say. But I
7 don't believe in throwing out the baby with the bath
8 water. We do have to finally decide what we're going to
9 do to make this a fair process so that Mr. Batra and
10 thousands and thousands of consumers like him do not have
11 to go through the process that he has gone through.
12 And by the way, when Mr. McCorkell was talking,
13 I was thinking, gosh --you know, I know that what he was
14 talking about was consumer lending. And I thought, boy,
15 I must have a really high credit score. I'm over 50.
16 I've been in my residence for --I own my residence.
17 I've been there over seven years. I only have a few
18 credit lines. I have low balances. I should have a very
19 high credit score.
20 But unfortunately I don't. And part of my --
21 well, it depends on what day I pull the credit. Because,
22 see, my ex-husband tends to marry women with names very
23 similar to mine. And I was wife three. My name is

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 117.
117 Page 118 119
118
1 Caryl. Wife four's name was Carmen, and Carmen's credit
2 always shows up on my report. And that's good old
3 Experian. I do have to tell you that is not Trans Union
4 and Equifax, but it does happen with Experian.
5 And by the way, I think we need to also
6 understand that there are three scoring models.
7 Everybody has heard FICO. And if I understand right by
8 what I've been told, Fair, Isaac does do the scoring
9 models for all three of these repositories that we're
10 most familiar with. One which is Equifax, the second
11 being Trans Union and the third being Experian, which we
12 all know was TRW.
13 Experian uses FICO. Equifax uses Beacon and
14 Trans Union uses Empirica. So when you get a credit
15 report --and by the way, all the credit reports I have
16 with me and a couple that I will talk about, are with the
17 okay from my clients, who said, go get them, Caryl.
18 (Laughter.)
19 MS. ISEMAN: Clients that are paying higher
20 interest rates than they should because the credit
21 bureaus wouldn't do their jobs. But you'll see that
22 there are three scoring models on each of these reports.
23 And as a mortgage broker, I get to pull the reports. I

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 118.
118 Page 119 120
119
1 get to see the credit scores. And, yes, my clients do,
2 too, because I give them a copy, even though the credit
3 scores are on there. And technically I understand that I
4 can do that.
5 So let me get to the issues of what I want to
6 cover, and I only have 15 minutes so I'm going to have to
7 talk fast. What you heard this morning was great, and I
8 really do believe that Fannie and Freddie are trying to
9 help the home buyers. But you know what? This morning
10 what you heard was the abstract. What you heard Mr.
11 Batra mention is the reality, and what I'm going to talk
12 about is the reality.
13 I'm going to talk about what was mentioned this
14 morning, number one, when they said, you know what? We
15 take anything through Desktop Underwriting. It doesn't
16 matter what the credit score is. You put it through and
17 we're going to try and get it through. Well, that's
18 great. The problem is, I'm a lender --I'm the mortgage
19 broker, and I have to go to the lender to do the Desktop
20 Underwriting.
21 Now, this is one of my lenders. This is the
22 form. This is my cover form that I extend through for
23 Desktop Underwriting. And one of the things that you're

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 119.
119 Page 120 121
120
1 going to see on here, the third box down says, FICO
2 score. Now, what they really mean is not a FICO score.
3 It's the middle of the three scores. And whether that be
4 Empirica or Beacon or the Fair, Isaac --either one.
5 Whichever is the middle of the three. Not the average,
6 but the middle of the three. That's what they want me to
7 put on there.
8 Now, right next to that this says, minimum 620.
9 Now, this was dated February of this year. That means if
10 you don't have a 620 medium score --or middle score --
11 don't send it to me. I'm not interested.
12 Now, here's my next one. This is the same
13 lender. By the way, this lender, I use them a lot for my
14 really good clients, because they have the best rates.
15 And I want my clients to have the best rates, and I
16 really, really would like to have most of my clients go
17 to them.
18 FICO score, the same thing, minimum, 640. So
19 from February to May, they said we don't want 620's any
20 more. You know why? Because this lender has good rates,
21 and they're going to skim off the top all the good loans
22 at the expense of the borrowers that have the 620 and the
23 630 and maybe even a little lesser score.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 120.
120 Page 121 122
121
1 And if you think that that's just one, I'm not
2 going to go through them all. But I have different
3 lender sheets --Chase and some other lenders here. I
4 think this is the best, though. Oh, I can't leave
5 Freddie Mac out, by the way. Freddie Mac has got a great
6 program. A hundred percent affordable goals. You heard
7 Freddie Mac say, oh, we want to look at them all. It's a
8 great loan. This is dedicated to low income borrowers,
9 minimum credit score, 660.
10 Okay. And then --oh, we don't deal with
11 professions or anything. Now, I understand --I spoke to
12 Fannie and Freddie today. I have to find it in here. I
13 know I'll find it. But if you're a doctor, please step
14 forward. First Union. Is there a doctor in the house?
15 if you're a doctor, they came out with a special loan for
16 you. You don't even have to have a 700 credit score.
17 You can do this with a 680. Or you can even be a D. O.
18 But we don't pay attention to any of these
19 things. Now --I mean, what you heard this morning,
20 let's look at everybody. We don't care. We don't pay
21 attention to minority people. We don't pay --we feel
22 everybody is on the same boat.
23 I didn't even realize this was going to be out

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 121.
121 Page 122 123
122
1 there today, but any of you that picked it up, turn to
2 page 27. On page 27 it says, who has riskier scores.
3 Now, what I have underlined here said, borrowers with
4 high risk scores particularly in light of various studies
5 indicating that minority families tend to experience
6 greater credit difficulties.
7 African American borrowers, for example, were
8 about three times as likely to have high risk credit
9 bureau scores, defined as FICO scores, below 620 as were
10 white borrowers, based on Freddie Mac's 1994 mortgage
11 purchases. And Hispanic borrowers were about twice as
12 likely as white borrowers to have high risk scores. But
13 excuse me. Let's ignore this. We're color blind. It
14 doesn't matter.
15 Let's get to talk a little bit about the
16 inaccuracies, because this is really where the problem
17 lies. Where you have mistakes on a credit report and the
18 borrower cannot get it off. And how prevalent are these
19 mistakes? I will tell you that of every ten credit
20 reports I pull, seven of them are incorrect. And the
21 other three we may not pay much attention to, because the
22 borrower had high credit scores on it and so we didn't go
23 through the report.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 122.
122 Page 123 124
123
1 But even when you do --and I have one here, and
2 my borrower's average mid score is like a 780. And even
3 he has two incorrect credit lines that are still showing
4 on there that showed --you know, that were paid off.
5 They're still showing balances. But he has such a high
6 credit score they don't care.
7 This was an article that appeared Friday, March
8 13, 1998. This was done by the Public Interest Research
9 Group. And again, I'm not going to go into the specifics
10 on the numbers here, but they point out all the
11 inaccuracies on the credit report. But there are two,
12 though, that I have to mention.
13 Nineteen percent --and these were the 133
14 consumer credit bureau reports. Nineteen percent
15 contained accounts that could not be identified or did
16 not belong to the consumer. Twenty six percent contained
17 credit accounts that had been closed by the consumer but
18 that were listed as open. And three percent had
19 incorrect Social Security numbers. That was a line that
20 was just total strangers. I know that feeling.
21 The inaccuracies are what effects the credit
22 scores. So that if I have mistakes on someone's credit
23 report that is pulling down their credit score, and they

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 123.
123 Page 124 125
124
1 have found this great house, and they have not 90 days to
2 close this transaction but 30 days to close that
3 transaction, it's only going to be by an act of God that
4 we're going to get it done. And my clients work with me.
5 God bless them. They follow everything I tell them.
6 They send return receipt requested letters.
7 I found phone numbers, by the way, where I
8 actually get a live person. Of course, you can't be sure
9 of what they're going to tell you. Because a gentleman
10 and I was on the phone --he didn't know I was on the
11 phone --from Equifax told my borrower --and this was
12 just a week ago --that Equifax had nothing to do with
13 the credit scores, that each creditor that submitted
14 their credit accounts decided the credit score. Of
15 course I had to hit the mute button at that point in time
16 and kind of let him know that I was listening and that he
17 needed to stop lying to my client.
18 I have to share with you where we get the
19 greatest inaccuracies and where it really hurts people
20 the most: medical, bankruptcies, medical collections,
21 people that have had catastrophic illnesses. Of course,
22 based on the predictability --that is, of course,
23 unpredictable and inconsistent. Based on this

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 124.
124 Page 125 126
125
1 predictability these people, because they've got these
2 low credit scores because they had to file bankruptcy,
3 out the window with them.
4 What do you mean? Because they had a
5 catastrophic illness, they're at 20 percent that their
6 share was maybe of their insurance policy, of maybe 50,
7 60 or a hundred or $200,000, and they could not pay it
8 and they were forced into bankruptcy. People that are
9 hard working people that had never paid a bill late in
10 their life ever. Never had a mortgage payment late if
11 they did own a home. They are so now unable to get the A
12 paper for a finance loan.
13 Now, can I get them a loan? Yes, I can get them
14 a loan at two percent or three percent higher than
15 anybody else. And I can get some of the stuff off when
16 they have filed that bankruptcy. Guess what? It still
17 shows as open as collections. And again, anyone who
18 wants to see this, I will show this.
19 I have one gal here, right in the process right
20 now of her loan. Her bankruptcy, which was partly
21 medical and partly not, was discharged 8/ 16/ 94. We are
22 in just about to be August 1999. She still has three
23 items that are showing up as unpaid collections. Guess

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 125.
125 Page 126 127
126
1 what happened to her credit score? Guess what happened
2 to the loan that we have to close in two weeks? I'm
3 getting her a loan, two and a half percent higher than
4 what she should be getting.
5 So all the savings that Fannie and Freddie spoke
6 to you about, that's great. But do you know at whose
7 expense those savings are? They are at the person who
8 could not get rid of his inquiries. They are at the
9 expense of the people that are going through catastrophic
10 illnesses and had to file medical bankruptcies.
11 And I want to give you another one. You know,
12 supposedly it's not only --are we supposed to look at
13 someone predicting the future and how they're going to
14 pay, because this is what we're looking at. Right
15 Fannie? You want to know you're going to get paid. And
16 the same thing with Freddie Mac. So they look at how
17 somebody paid their previous bills.
18 Well, what about with downsizing. How many
19 people have been put out of work? I know we had a lot of
20 that happening in San Diego. I know that even in this
21 great economy, you have people that get unemployed, and
22 they may have to go for three or four months where they
23 become a little slow paying. But for the 30 years

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 126.
126 Page 127 128
127
1 before, they never had a late payment, ever. So for
2 three or four months, they have a car payment that's
3 maybe 60 days later or whatever. Are these people bad
4 risks if they want to go and buy a house right now? The
5 guy is back working.
6 I have that. I have an airline pilot. And by
7 the way, not in San Diego, in Michigan. He was out of
8 work. He flew for 30 years. They have a great credit
9 record, because as Mr. McCorkell told you, all the good
10 stuff does remain. It stays on there. You look at it.
11 They had a house. I mean everything. Never a late
12 payment. He was out of work for six months. They had a
13 60 day car payment --60 days late on a car payment and
14 two credit cards. Two credit cards that had 30 day
15 lates.
16 And I have to tell you where their credit scores
17 are, because it's recent. Anything recent takes really
18 from that credit, and they have to do a refinance on
19 another property that they own. We're at a very, very
20 low loan to value. But because it's nonowner occupied
21 and they want to do the loan to fix the house up because
22 the tenant just died, I have to tell you, they're a B
23 paper loan. And should they be 54 percent loan to value,

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 127.
127 Page 128 129
128
1 someone who for 30 years never missed a late payment?
2 But the predictability, these people are going
3 to be lousy borrowers. They're not going to pay back,
4 because they have a 610, I think from one of the bureaus,
5 and a 600 --oh, no. They're not going to make the
6 payback. That's why I have difficulty with the setup
7 with the credit scoring.
8 Now, I said before, do we throw the baby out
9 with the bath water? No. Could credit scoring be a tool
10 and a useful tool? Absolutely. Because people who
11 really don't fall into any problems and keep making their
12 payments on time, they have a really, really good payment
13 history. But I don't need a statistic to tell me that.
14 Any underwriter could look at that type of file and say,
15 hey, I can approve it. I don't need someone to give me a
16 credit score.
17 But let's say so they have really good credit
18 scores, and because of that we can move real quickly.
19 Well, that's great, because that person maybe should be
20 compensated and have really a loan move through the
21 system quickly and should have a little better interest
22 rate. I have no problem with it.
23 But the process as it is right now has got to be

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 128.
128 Page 129 130
129
1 changed. The scoring models have to allow for some
2 flexibility. It has to take into account medical
3 bankruptcies or that type of concern, or the short term
4 unemployment. There's got to be some flexibility. Right
5 now there is absolutely none, and that's why I have a
6 problem with the system.
7 These are my suggestions --because I know I
8 have to cut myself short. Number one, everyone in this
9 room that is either representing a regulatory or a credit
10 agency, or represents a lender, or whoever you represent,
11 if you are part of this process, we must work together to
12 get this worked out.
13 Now, there are three bureaus that are going to
14 go kicking and screaming to work this out, and that's
15 Trans Union, Equifax --and I know there's a gentleman
16 here who is representing the three bureaus. I don't know
17 if anyone from those --from the three repositories are
18 here. But you've got to clean up your act. It shouldn't
19 take 30 days for a bureau --for a mistake --an
20 inaccuracy --on a credit report to be done, not in this
21 day and age of technology.
22 And also creditors have to start reporting more
23 promptly and correctly. And when they increase someone's

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 129.
129 Page 130 131
130
1 credit balance so that there will be a greater disparity
2 between what the client owes and what their balance is,
3 they have to report that, and it has to be not three
4 months old. It has to be current, so that we're on par
5 with everything.
6 And inquiries, unless --not only if somebody
7 even checks for an inquiry and, you know, was really
8 sincerely checking on it. But only --only --if an
9 account came as a result of that should that be counted.
10 And why do we deal with this mortgage thing and 30 days
11 of shopping for a car? If someone is shopping for a
12 mortgage, isn't it their entitlement to get the best rate
13 and find the best mortgage? And it might take them 60
14 days and it may take them a hundred calls.
15 And they should have the right to do that and
16 not be penalized, which they are now. If it takes them
17 more than 60 days and it carries over, and they have five
18 or six new inquiries, guess what. Their credit score
19 goes down for trying to do for themselves the best job.
20 And the same thing with credit cards. Don't you
21 want to get the best interest rate on your credit cards?
22 Shouldn't you have the right to shop around. Well, don't
23 do it anywhere near the time that you're thinking of

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 130.
130 Page 131 132
131
1 doing a refinance or buying a home. And you should have
2 the right, and you should have the right to consolidate
3 all your monies. You should have the right to do that,
4 and only if a new account results as an effect of that,
5 that is the only time that inquiry should be counted.
6 And I think we all really need to start paying
7 attention to this. I know we're going to have a question
8 and answer session, and I know that a lot of other people
9 have things to say, so I'm going to thank you all for
10 listening to what I've had to say. I'll be glad to
11 answer any questions when they come up later.
12 Thank you.
13 (Applause.)
14 MS. TWOHIG: Thank you very much, Caryl. Our
15 next panelist is Marcia Griffin, who is founder and
16 President of Home Free USA, and the former Vice President
17 of a mortgage servicing firm. Home Free USA is a HUD
18 approved, nonprofit membership organization that promotes
19 home ownership and assists perspective home buyers with
20 such things as education and counseling on the home
21 buying process, credit counseling and repair, and offers
22 other guidance.
23 And we welcome Marcia to give us some comments

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 131.
131 Page 132 133
132
1 also.
2 MS. GRIFFIN: Thank you. Thank you. Well, I'm
3 going to just give a different perspective here, if I
4 might. I represent hundreds and thousands. Last year we
5 interacted with over 10,000 perspective homeowners. One
6 of the issues that I think that we're --as we discuss
7 credit scoring today, I want you to know that we use
8 credit scoring every day, and we use it in the first
9 meeting that we have with every homeowner --every
10 perspective homeowner.
11 One of the issues that I would like to also
12 bring up is that through education of the credit scoring
13 issue, education of how lenders work, what Freddie and
14 Fannie are looking for, the education is key. Education
15 to the mortgage industry. Education to the nonprofits.
16 And education to the perspective homeowners, which is
17 what we're attempting to do here.
18 We have had good and bad results with the credit
19 score. You know --and I must say, though, and let me
20 just give you a little bit of sort of background. In
21 Home Free we do a lot of education. We have anywhere
22 between eight and 21 classes a month, because people need
23 to know. They need to know their credit score, and they

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 132.
132 Page 133 134
133
1 need to know it early from our advantage point. They
2 need to know what they need to do, how their credit needs
3 to be cleaned up. They need to know the good, the bad
4 and the ugly before they go to Caryl.
5 From our advantage point, if people are
6 prepared, they will get approved. In Home Free we have a
7 100 percent approval rate. Every single person that has
8 been through our process. Now, let me just tell you,
9 some people can fly. Others have to walk, crawl or take
10 a boat or train. I mean, it really varies in terms of
11 the time that we spend with them.
12 But the reality is that from our perspective,
13 and through the perspective of our partners --Freddie
14 Mac has been a wonderful partner. We've learned a lot
15 through credit scoring with them. Chevy Chase Bank has
16 been a great partner. We're in partnership with other
17 lenders.
18 But the reality is that it is important that
19 people understand where a credit score positions them and
20 what they need to do. Now, from the education
21 standpoint, the reality is that a lot of the nonprofits -
22 -and we're in the counseling mode --don't --we're not
23 really in sync on what people need to do to make

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 133.
133 Page 134 135
134
1 themselves better, even if it's going to be, you know, a
2 six month period with the nonprofit. This is where the
3 education of the nonprofits need to occur.
4 But in this class that I conducted --in fact, I
5 certainly open an invitation to any of you. I have this
6 class every Thursday night at 6: 30 at my office here in
7 Washington. It's called Ten Home Buying Secrets Everyone
8 Ought to Know.
9 And the first secret is that the mortgage
10 process has changed, that people should not think about
11 home ownership and just, you know, jump out here, go to a
12 realtor, sign a contract and go to lender. They don't
13 know where they stand financially. You know, they know
14 that their credit situation might not be the best. But
15 there is no preparation at all.
16 So from our advantage point, the process has
17 changed. They need to get prepared. People need to get
18 prepared before they go to a lender so that they can be
19 assured of getting the best rates --the most money,
20 because that --the money issue I know we're not talking
21 about this today. But I can tell you, in Home Free USA,
22 65 percent of our people have credit scores of over 620.
23 We deal only with under served populations: minorities,

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 134.
134 Page 135 136
135
1 African Americans, Latino populations, immigrant
2 populations and also whites.
3 So the credit score issue is one factor.
4 Clearly, the other factor, certainly in our environment,
5 is that we have people with a 700 credit score and no
6 cash. I mean, this is a big issue that we've got to deal
7 with.
8 But let me just get back to my class for a
9 minute. So the first part --the first secret is that
10 the process has changed. And the second secret is to get
11 your credit score. Get your credit report immediately,
12 not from the credit reporting agencies, because when you
13 write, they don't give you your credit score. They need
14 to really get their credit report with a credit score, so
15 that they can understand where they are. And they need
16 to meet with an organization like Home Free, like the
17 NHS's and, you know, other nonprofits around, and perhaps
18 even lenders, to examine that credit report and begin
19 making these changes early on.
20 Clearly, where Fair, Isaac, Freddie and Fannie
21 need to come to the table, we need to clearly understand
22 what the impact is if, you know, a person's credit score
23 --let's say --and I've heard several things from Fair,

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 135.
135 Page 136 137
136
1 Isaac. You can't have, you know, too many credit cards,
2 and they can't be up to the limit.
3 Well, you know, we need to have some clear
4 factors. In fact, I would love to have someone to
5 produce just a little report, something like 30 essential
6 things to know about your credit score. Things that you
7 --I mean, and basically how you can improve it. You
8 know, what's causing the problems. I mean, there seems
9 to be a little bit of question about this across the
10 board.
11 So from our advantage point, clearly education
12 is the key. We use the credit score at the very
13 beginning in what we call our mortgage profile. I mean,
14 that is the first one on one meeting that we have with
15 the person. We pull a three merge credit report, and we
16 see the credit score, and we work with that. And we have
17 seen that, you know, when they go to the lender, that
18 credit score, you know, has improved.
19 The issues, though, are that, you know, what's
20 going to make it improve significantly and what's holding
21 it back. It's a very interesting part that you brought
22 up here, Caryl, about, you know, the inaccuracies. Well,
23 we know the inaccuracies with a credit report. But, you

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 136.
136 Page 137 138
137
1 know, we haven't run across, you know, many issues that
2 sort of relate to how we could tell --I mean, the
3 question about the scoring. I mean, you can't really --
4 no one has a feel for how these scores --how they even
5 come up with the scores. So it's difficult to really
6 assess whether the score is right or wrong or just
7 whatever.
8 But I can tell you that, you know, there are
9 people who certainly when they have a credit score of 500
10 --when we see a credit score of in the 400's, we tell
11 them to just stop. There is nothing more --stop. Hold
12 it. Let us take your hand and walk you through a
13 process. Because while there may be some errors there,
14 we doubt if the credit score is showing up at 490 that
15 it's going to --there is enough error in there to make
16 it a 700 credit score.
17 Where we are finding some issues, however, is
18 right now we have --one of our members has a 710 credit
19 score. Now, this is pretty good, you know. She has a
20 710 credit score, and the lender is having question after
21 question after question about this loan. Clearly --we
22 talked earlier about the Desktop Underwriter and the Loan
23 Prospector. We have asked Freddie Mac to give us the

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 137.
137 Page 138 139
138
1 Loan Prospector, because we have found that when several
2 of our members have gone to lenders and they've begun
3 questioning certain aspects of whether that loan should
4 be approved, then they've gone --we've sent them to
5 another lender and the loan has gotten approved using
6 these technological methods here with Loan Prospector and
7 Desktop Underwriter.
8 I'm sort of betwixt and between, as I said, you
9 know, as I represent under served people, as I represent
10 home buyers --future home buyers --and as far as the
11 credit scoring issue, I think it's a great tool. Quite
12 frankly, I think we need to be better informed and better
13 educated about what's going to make the difference in
14 enhancing that credit score in our environment.
15 At Home Free USA we have on average between 45
16 and 150 people per class. I mean, this is --and we have
17 three classes a week. People have a thirst for
18 knowledge. Under served people in this country are
19 trying to do better. We preach to them, through churches
20 and through employers and through, you know, unions and
21 the like, that look, we have got to go into the next
22 millennium better off than we have been. People want to
23 be informed. They want to get their credit score up.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 138.
138 Page 139 140
139
1 But we just need to have consensus about what
2 it's going to take and what people need to do. And it
3 doesn't need to change every five minutes. But what they
4 need to do and basically how we can all be on the same
5 level so that we can increase home ownership. People
6 want to become homeowners.
7 And, you know, there are a lot of people that
8 are pretty bad off, but even those people --you know,
9 what we tell our people is that despite where you are
10 today, you can be better off tomorrow. A misty morning
11 does not signify a cloudy day. You come on and we're
12 going to work with you with that credit score, with the
13 credit, with these statements and everything else.
14 You know, it is from that advantage point that I
15 sort of represent other nonprofits who work with the
16 people, who if we can get this thing together, by the
17 time the people get to Caryl and all of the mortgage
18 lenders in here, the loans will be right. We won't have
19 to send them to B, C and D paper lenders, which I think
20 is the worse thing of the world.
21 And certainly when they get to Freddie and
22 Fannie through counselling, through education and also
23 pulse purchase counselling --you need to stay with these

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 139.
139 Page 140 141
140
1 people for at least two years --at least at Home Free
2 this is what we do --and keep them with you and teach
3 them what they need to do to better themselves
4 financially. Everybody wants to be better, and this is
5 what the objective of, I think, all of us would be.
6 Thank you very much.
7 (Applause.)
8 MS. TWOHIG: Thank you very much, Marcia. We're
9 running a little late, but as promised we would like to
10 give the audience a chance to raise some questions and
11 issues so that we can be sure to cover them this
12 afternoon.
13 So why don't we take a few minutes and do that
14 now. And there are hand held microphones, so if you
15 raise your hand you can be given the microphone. And
16 those of you in the overflow room, if you have questions
17 at this point, if you come to the main room we can try to
18 get your questions and comments in.
19 MR. PREUDHOMME: Hi, can you hear me?
20 MS. TWOHIG: Yes.
21 MR. PREUDHOMME: Okay.
22 MS. TWOHIG: And please first identify yourself
23 and identify any affiliation you have.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 140.
140 Page 141 142
141
1 MR. PREUDHOMME: Yes. Hi. My name is Bruce
2 Preudhomme. I'm here not only representing myself and my
3 family --okay. Somebody wants the spelling of my name.
4 It's Bruce Preudhomme. That's P-R-E-U-D-H-O-M-M-E.
5 As I said, I'm not only representing myself and
6 my family, I'm also representing two consumer activist
7 organizations I work closely with. One is called the
8 Victims of Chevy Chase Bank. Sorry.
9 (Laughter.)
10 MS. TWOHIG: Okay.
11 MR. PREUDHOMME: And the second is the Economic
12 Justice Alliance. Okay. In both cases what we're
13 dealing with has to do with credit scoring, credit
14 reporting and problems centered around that whole arena.
15 The biggest problem I see --well, there's really two big
16 problems. The data, as has already been mentioned. In
17 my field of computers, we have a term called garbage in,
18 garbage out.
19 Okay. And we've got that going on big time in
20 the credit bureaus. The data they get in is garbage. It
21 comes from the lender. It can actually be used as a tool
22 to attack somebody that has been a vocal person. I for
23 one --after closing an account I paid off, I had a

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 141.
141 Page 142 143
142
1 lender a year after it was closed, report me as 60 days
2 delinquent, which I had to battle and go through the
3 whole 90 day process. I just got my letter the other day
4 saying okay, yeah, you were right. It's not that. And
5 then they put in some other verbiage, okay, which is
6 another pack of lies that I have to battle.
7 The data as far as inquiries is totally bogus,
8 too. Back in 96 I had my credit rate increased on my
9 gold MasterCard from 18 percent to 25.99 percent. Not
10 because I was paying late. Not because I was paying
11 slow. But because of information contained in my credit
12 bureau. And one of the big factors was the number of
13 inquiries.
14 Well, I checked. The number of inquiries that I
15 had during the last year was nine, okay. It happened to
16 be that year I bought two car phones. And because I run
17 an on-line site where I dial up access, I got another
18 four phone lines in the house. Most of those credit
19 inquiries were for phones. Now, I don't know why that
20 would affect me. Why that would justify Chevy Chase Bank
21 raising my interest rate, but they did, okay.
22 I battled this and in --you know, it took I
23 guess two years to go through the process, trying to go

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 142.
142 Page 143 144
143
1 through the system. And, you know, in every step, you
2 know, I contacted the FTC. I contacted the OTS. You
3 know, they didn't want to deal with it. They would
4 rather just say, oh, well, you know, do a better job at
5 paying your bills. I pay them off, you know, so that
6 wasn't the problem. Then they say, okay, well, maybe you
7 should get an attorney. And I have a real problem that I
8 have to get an attorney to get the laws in this country
9 enforced, and to get the laws in this country done right.
10 Now, back to what I was saying, there are two
11 basic problems. Data, as I said, is one. The other is
12 the whole credit system. We've got the cart before the
13 horse. We have lenders funneling garbage data into the
14 credit bureaus. Instead, that data should come to the
15 consumer first. You know, the consumer should have the
16 right to say, hey, wait a minute, this is totally bogus.
17 I should be granted permission to say hey, this
18 never happened before it gets posted. I should get a 60,
19 90 or 120 day period where I can clean it up before it
20 gets posted, rather than waiting until after it's posted
21 and I happen to stumble upon it, you know, and that's the
22 way it is right now.
23 So basically one of the premises that this

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 143.
143 Page 144 145
144
1 country was founded on that you're innocent until proven
2 guilty, the reverse is happening. You are presumed
3 guilty, okay. There's a lot more that I could say, but I
4 want to give other people a chance to talk.
5 MS. TWOHIG: That would be great.
6 MR. PREUDHOMME: Thank you for listening.
7 (Laughter.)
8 MS. TWOHIG: Are there any --no, we appreciate
9 your comments, but it would be good if people could
10 somewhat limit their comments. Does anyone else want to
11 raise questions or make a comment at this point? Again,
12 we're just trying to get issues out on the table so we
13 can maybe get some responses and answers this afternoon.
14 MR. LE FEBVRE: Do you need my name and
15 everything first?
16 MS. TWOHIG: Yes. Please identify yourself.
17 MR. LE FEBVRE: Richard Le Febvre. Affiliation
18 is AAA American Credit Bureau. I'm here for --
19 MS. TWOHIG: Oh, could you spell your name for
20 the record?
21 MR. LE FEBVRE: Le Febvre, L-E-F-E-B-V-R-E.
22 We're a mortgage credit reporting bureau, and we service
23 most of the United States. And in listening to Fannie

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 144.
144 Page 145 146
145
1 and Freddie and Fair, Isaac, we're --and the
2 repositories --we're in the trenches every single day.
3 When a consumer goes to Caryl or anybody else that is a
4 mortgage broker, we're actually pulling the report and
5 actually analyzing it and seeing it on a daily basis.
6 And, of course, the repositories take a lot of
7 the blame. Every consumer group from the legal side to
8 the consumer groups blame them. But a lot of the times,
9 in my own personal experience and in the reports that we
10 do each and every day, it's the creditor that's reporting
11 that data. And one of the things that I feel real
12 strongly about is back in 72 when the Congress enacted
13 the Fair Credit Reporting Act and then amended it again
14 in 97, they knew at that particular time with no blame
15 that the data was wrong. There were creditors
16 misreporting and there are certain things are there.
17 And then DU and LP come along. And my biggest
18 concern is if Congress passed the Fair Credit Reporting
19 Act knowing there were inaccuracies and gave consumers
20 certain rights, why does Fannie and Freddie don't think
21 that they have to give that consumer those same rights as
22 it stands right now.
23 In the old days what would happen if Mr. Batra

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 145.
145 Page 146 147
146
1 had gotten his credit report from us, a credit report is
2 generated. He's sent a consumer copy. He then has the
3 right to dispute that data. Now, because of the 30 day
4 turnaround, the repositories cannot get that data
5 verified that quickly. That's why mortgage credit
6 reporting companies came into effect, because they can
7 represent or protect both consumer and lender to verify
8 that data when Mr. Consumer says it's inaccurate.
9 Now, in the system --Fannie Mae, to get on
10 their system, comes to me and asks for $75,000 if I
11 could, but I can't. Then they ask for transaction fees
12 to come through a system that they don't give the
13 consumer the same right. My proposal, and one of my
14 biggest issues is, if that data is wrong, Fannie and
15 Freddie must then allow third party intervention, either
16 by the repository level or by a mortgage credit reporting
17 company that's out as a third party to verify that data
18 to show it.
19 A lot of credit --now that Fair, Isaac uses a
20 scoring model, a lot of creditors --and if you talk to
21 the repositories, all creditors don't report the same
22 way. High credit to one creditor is different than high
23 credit to another. And if that model is looking at the

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 146.
146 Page 147 148
147
1 balance --the high income ratio, and if you have a
2 Discover Card or an American Express card, every single
3 month you're at your limit, even though you're truly not.
4 And there must be a mechanism in place that both
5 Fannie and Freddie will allow third party intervention
6 and to then correct that file and then send it back
7 through. LP and DU is a great operating system. One, I
8 don't think I need to pay to get on their system because
9 mortgage companies like to use my company or Credco --I
10 know they're represented here --and numerous other
11 companies. But there has to be a way that the consumer
12 and both Fannie and Freddie will then allow third party
13 intervention, and then accept that data, which in turn
14 then will recalculate the score.
15 Right now to talk to Equifax --Equifax right
16 now, out of the three national repositories, has what
17 they call an expedite program. That means, I'm a
18 mortgage credit reporting company, and if a consumer or
19 Caryl or anybody else in this room has inaccurate data
20 that we're able to prove, they then send me a document.
21 We at the mortgage credit reporting company authenticate
22 that document, send it to Equifax, and Equifax is
23 committed to correcting the file within 24 to 48 hours.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 147.
147 Page 148 149
148
1 Which in turn --then we pull the file, which in turn
2 recalculates the score.
3 The other two bureaus have not done it yet. And
4 I think if the repositories working together with
5 mortgage credit reporting companies --and I'm sure there
6 has to be a code of ethics. There has to be certain
7 fraudulently --something in place that they don't
8 receive fraudulently done documents.
9 But my proposal would be, if the document is
10 then faxed from the consumer --if Mr. Batra was able to
11 verify that Discover Card misreported his credit card and
12 received a letter from Discover Card, then it would be
13 sent to my company. We would authenticate that letter,
14 by calling the person --the author of that letter --and
15 then forwarding that on to the repositories, and having a
16 relationship between the repositories and us, so that
17 data they feel comfortable correcting.
18 And that's the system that needs to be in place,
19 because when you talk to everybody, they're upset about
20 the credit scoring system. They're upset with the
21 repositories. And I'll be honest with you, the
22 repositories --all they are is an acceptance of data.
23 It is impossible for them to turn around and look at

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 148.
148 Page 149 150
149
1 every trade line that's reported to them every single
2 month and know if it's right or wrong.
3 The only way the repositories know if it's right
4 or wrong is if the creditor corrects the file, either by
5 magnetic tape or any other means, or if the consumer
6 disputes. So a lot of the pressure is put on the
7 repositories, and they're not the problem.
8 We had a file a couple of weeks ago where we
9 were dealing with BankOne, and BankOne reported a
10 gentleman's perfect BankOne account as a Chapter 7
11 bankruptcy. It decreased his scores by over 70 points.
12 And the only one of the repositories --in dealing with
13 BankOne, he received a dispute letter that said it would
14 be corrected in 30 days. This is now 90 days later and
15 it still hasn't been corrected.
16 At the same time, three of his credit card
17 vendors had notified him that the increase in his credit
18 report went from 9.9 to 22 and a half. And he never had
19 a Chapter 7 bankruptcy at all. He always paid as agreed.
20 So the repositories themselves, I believe, need to lead
21 the front in dealing with third party intervention to get
22 this stuff.
23 And one other quick example, and my biggest

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 149.
149 Page 150 151
150
1 concern, is discrimination towards Hispanic or minority,
2 blacks, whites or whatever it may be. We did a file the
3 other day, and I analyzed it. And a file was pulled by
4 my company on 6/ 7 and then it went through the DU system
5 on 7/ 9, which is one month and two days later.
6 I analyzed the raw credit from all three
7 repositories. Equifax had four trade lines, TU had five
8 and Experian had seven. The only thing this gentleman
9 did was miss a $20 monthly payment to Discover. And of
10 course he disputed it. He said he was not late. But
11 Experian increased his --not Experian. Fair, Isaac
12 showed his score 76 points higher with Experian, 151
13 deduction with TU, and 178 deduction with Equifax. But
14 he paid his mortgage --and he only had one derogatory
15 item on his credit file, and it was a $20 payment. And
16 he's had a mortgage history for three and a half years
17 that was paid on time.
18 So my analysis of it all, since we're in the
19 trenches each and every day, is that something needs to
20 be done, and it needs to be done drastically. Because
21 right now about 80 percent of our incoming phone calls
22 are why is my score this, and how do I increase it, and
23 so on and so forth.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 150.
150 Page 151 152
151
1 Thank you.
2 MS. TWOHIG: Thank you very much for your
3 comments. I think it's important as we think about these
4 issues to think of something --I think Pete McCorkell
5 first suggested this afternoon. That we also have to
6 think of what issues are we talking about that have to do
7 with credit scoring, and what are just the credit
8 processes that would also exist in a judgmental decision.
9 As I understand it, your first comment that you
10 made did have to do with the speed of automation, that
11 makes it more difficult for the consumer, necessarily in
12 the course of that decision making process, maybe at that
13 same time to dispute errors.
14 Are there other questions or comments? Yeah.
15 MR. SMITH: I'm Don Smith. I'm the President of
16 Great Arizona Mortgage Corp. in Scottsdale, Arizona. And
17 this is my first visit to Washington, and what a great
18 town. And I want to thank you for having me at the
19 hearing today.
20 With regards to why I'm here, those two loans
21 that Richard just brought up were loans that were done in
22 my office. The second one was a Hispanic couple that had
23 700 and some change in their FICO scores, and they're now

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 151.
151 Page 152 153
152
1 500 and something in less than 30 days, and there is no
2 difference in the file. Try to figure that out and
3 explain that to a borrower that barely speaks English,
4 why when I told him 30 days ago he would have no problem
5 whatsoever. He's made arrangements for his move. And
6 before I came out here, I had to look him in the eye and
7 say, we may be having a problem.
8 Now, of course, if you listen to Fannie Mae and
9 Freddie Mac, they explained to you that there are ways
10 that we can go about this. We can expedite it. There
11 are no changes on the report. The numbers have shifted.
12 As a lender --I've been doing this 19 years as a
13 mortgage lender. I've been a realtor for 22 years. I
14 have done thousands of FHA loans, VAs and conventionals.
15 And I love what we've done with the automated
16 system. Unfortunately, the premise that this is based
17 upon is based upon a perfect world or in a vacuum. We
18 are not living in a perfect world. And we need to take
19 the time as we talked about --you know, let's talk about
20 what makes up the scoring system with Fair, Isaac. Let's
21 talk about if you have a $1,500 limit on your credit
22 card, that it says you only owe $200 and it's a $1,500
23 limit.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 152.
152 Page 153 154
153
1 But we've got numerous credit reports that I'm
2 sure that everybody in the room has seen over and over,
3 where it says high balance is zero and you owe $1,500.
4 How do they let you do that? I want to know how you do
5 that. How do you borrow $1,500 more than your high
6 limit? It's calculating under scores. And if it's not,
7 I want somebody here to tell me that that's not how it's
8 being done, so I can go back and tell my clients, the
9 ones that I see on a consistent basis, that look, this is
10 how we work it.
11 This is what's important. Instead of saying
12 well, you know, if you pay your bills on time, if you go
13 to church every week and you pay tithe, you do this and
14 you do that, God is going to take good care of you and
15 don't worry about Fair, Isaac. That is not the case.
16 What we need to take care of, is let's kind of get down
17 to it. Let's crunch the nuts and the bolts of this issue
18 and come up with it and say --just like we're talking
19 here with Simi. I don't know. I don't know.
20 You know, it just comes across the paper, and by
21 god, you're either going to get approved or you're not.
22 Unless you take it to an underwriter that knows you and
23 trusts you, and she's willing to step out on a limb and

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 153.
153 Page 154 155
154
1 say, well, I don't give a damn if they're going to buy it
2 or not. You know what? This loan is worth underwriting,
3 and we're going to make the loan.
4 The biggest issue that I've got still is the
5 reporting issues that we have from banks. By the way,
6 that one file, the first one that he mentioned to you,
7 the borrower was buying a $300,000 house in north
8 Scottsdale. It was up to $400,000 before he closed. I
9 had approved him. He had great credit overall going into
10 the transaction.
11 We're getting ready to close on the deal. I
12 said come on back in the office. Let's take a look at
13 it. BankOne --and I'll tell you the name of it, because
14 I've got all the documentation here with me. BankOne
15 rates him on his credit report now as being six times 90
16 days late and bankrupt. His credit score is in the
17 porcelain convenience. The man is livid. His wife had a
18 miscarriage over this period of time. They did not close
19 on the house.
20 I had him come in. I have a letter in the file
21 that I was asked to give you a copy of so you could
22 review it and take a look at it. I have the
23 documentation. In their infinite wisdom, BankOne, when

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 154.
154 Page 155 156
155
1 we called them on the telephone and explained to them
2 that --and we got the automated first. We did the
3 1-800, wait in line and die.
4 (Laughter.)
5 MR. SMITH: We did that and got that
6 information. And you know what it said to me? Never
7 been late. Well, okay, fine. So I hit the O button, the
8 one you're supposed to keep hitting until you get a live
9 person. They got on the telephone. She says, well, Mr.
10 Smith --and I told her who I was and that the client was
11 sitting there. It's a mutual client of ours. And
12 explained to her, your system says that he has paid right
13 on time. She said that's absolutely correct.
14 And I said well, why is it being reported on the
15 credit bureau, and she goes, I have no idea. And I said
16 well, what's your position. Well, I'm the supervisor.
17 How long have you been doing this? Well, what's it to
18 you, is how she came back to me. And I said no, I'm
19 serious. How long have you been doing this? Are you
20 familiar with the Fair Credit Reporting Act? And she
21 says well, yes. And I said are you familiar with it,
22 like have you read it? And she said well, yeah, I'm
23 familiar with it.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 155.
155 Page 156 157
156
1 And I said well, you have an obligation. You
2 need to get me a piece of documentation. We are trying
3 to close on this man's home. I have less than two weeks.
4 This is a surprise. You've already told me that it's not
5 his information. You need to send this information out
6 to me right now. I said, here's his name. You have it.
7 Here's his Social Security number. You need to reference
8 the account. And I need that letter, and you need to fax
9 it to me immediately. I'll give you to Friday. This is
10 Wednesday. It should be easy for you to do. I know you
11 feel bad.
12 (Laughter.)
13 MR. SMITH: I got the letter Friday. And I have
14 a copy of the letter, and I'll be glad to show it to you,
15 because this is what goes on in the real world out there
16 for many of these people out here that are doing these.
17 The letter says: Dear Mr. Smith. Reference the
18 account. Please be apprised we have removed this line
19 item from your credit bureau. For mine. They didn't
20 take it off his. They took it off of mine.
21 (Laughter.)
22 MR. SMITH: She doesn't even know my Social
23 Security number, but I'm sure that all Don Smith's don't

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 156.
156 Page 157 158
157
1 have a BankOne account number. Well, it gets better. I
2 called her back up. I was on that terminal hold thing
3 again. Okay. My client's in the office. He's a big
4 guy. Don't make him mad. Thank God he's a friend. So
5 we're standing there waiting, you know, on hold.
6 We get them back on the line. Noel Simpson, how
7 are you? Who is this? It's Don Smith again. Oh. I
8 noticed the letter that you wrote me. You know, it was a
9 pretty good written letter, but I've got a couple of
10 items I would like to bring up to your attention. You've
11 left out Mr. Acton's name and Social Security number, and
12 you just said that you removed the item.
13 We don't want the item removed. It's his
14 credit. He's had it. He's been paying on it. You've
15 told me that. I want you to take it and state that he's
16 never been late, and he's never been bankrupt, and we are
17 sorry and apologetic. Well, that's going to take a
18 little while. No, Noel, we've been through this. Let's
19 try this again. Get it taken care of.
20 She sent me out another letter. She removed it
21 completely off his credit report. There is no trade
22 line. It doesn't show any payment history. It doesn't
23 show any balance. So we can't figure out if he still

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 157.
157 Page 158 159
158
1 owes the money or not. But guess what's still in the
2 column? He's still showing bankrupt. His credit scores
3 have depreciated and depleted over this period of time.
4 He is now in the mid-500's, when he started --I mean,
5 he's dropped over a hundred points. The man lost over a
6 $100,000 of equity on the house.
7 And I'm going to shut down right now. I see you
8 all looking at the clock. But this is what the rest of
9 us are doing business with out there. This is what we
10 need your help on. Let's put some teeth in the FTC, if
11 you're going to have issues for the credit bureaus with
12 what they have to do, and for the people that provide the
13 information.
14 Thank you very much, and you all have a nice
15 day. A great city.
16 (Laughter.)
17 MS. TWOHIG: You know, when we announced this
18 conference and we started getting feedback, we knew we
19 were going to get some lively comments, but I think we
20 under estimated how lively.
21 Any other quick comments before we break?
22 MR. EVANS: Right here.
23 MS. TWOHIG: Mike?

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 158.
158 Page 159 160
159
1 MR. EVANS: I'm Mike Evans, Fair Lending
2 Specialist with the FDIC. And I noticed this morning
3 that the individuals with Fair, Isaac and some other
4 individuals dealing with credit scoring have said how
5 credit scoring helps fair lending.
6 Well, I want the issue addressed. How does it
7 help fair lending when you look at the data from the Home
8 Mortgage Disclosure Act that is released every year. The
9 denial rates for minorities continue to increase every
10 year. And I also noticed that in the data that will be
11 released in the next couple of days, if you look at
12 closer, denial rates for whites are increasing, also.
13 MR. MEDINE: I think, Mike, that's an excellent
14 question, and we're going to have a panel on fairness
15 after the break, and why don't we have that as the first
16 topic during that panel.
17 MS. TWOHIG: We're going to need to --okay?
18 MS. PRATT: Hi. Sarah Pratt, now with the
19 National Fair Housing Alliance. I'll be very short. It
20 seems to me that there are a couple of really important
21 questions that we need to continue to deal with.
22 One is a synapsis of the issues raised earlier.
23 And that is, what difference --what are the sources of

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 159.
159 Page 160 161
160
1 the various items in the credit scoring technique? How
2 are things weighed? What difference does it make if you
3 have a trade line from Household Finance versus a
4 MasterCard, and why should it make a difference? Where
5 do these scores come from? That's the first item, and we
6 don't know the answers to those questions.
7 The second question it seems to me that ought to
8 be talked about some is the relationship of credit
9 scoring to the availability of loan products, and in
10 particular to referrals to the sub-prime market. In
11 other words, where --what difference does it make if you
12 have a particular credit score as to what loans are
13 actually available and are the most favorable loan
14 products available to you based on your scores? Why not,
15 if not? And at what point do you end up in the sub-prime
16 market because you have a particular score, and what
17 impact does that have on your finances for the future?
18 And even what impact does it have on your future credit
19 scores?
20 Thank you.
21 MS. TWOHIG: Thanks, Sarah. One or two more
22 questions or comments, and I think we'll need to break
23 for lunch.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 160.
160 Page 161 162
161
1 MR. FISHER: It will be quick. Greg Fisher,
2 creditscoring. com. I've listed 25 questions on the web
3 site, creditscoring. com. And that's G-R-E-G,
4 F-I-S-H-E-R. And I have copies of them here, if you
5 would like them. But I would like to ask the first five
6 here.
7 What percentage of loans evaluated by Fannie Mae
8 and Freddie Mac automated mortgage underwriting use
9 scores created by Fair, Isaac? Will the credit reporting
10 agencies release data to anyone who wants to create a new
11 scoring system? If I'm applying for a mortgage, how much
12 does it cost to have a score updated immediately, if I
13 find errors on my report? Do you allow creditors to see
14 scores for the purpose of evaluating existing customers?
15 Do credit card companies access credit scores months or
16 years after issuing the card to determine the risk of a
17 cardholder to raise or lower his rate? And the fifth one
18 out of the 25, how can I tell how many credit cards to
19 have to avoid that factor lowering my score?
20 MS. TWOHIG: Thank you for those questions.
21 There was someone else, I think, who had their hand up.
22 Yes?
23 MR. MANALP: My name is Yuksel Manalp.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 161.
161 Page 162 163
162
1 MALE SPEAKER: Spelling?
2 (Laughter.)
3 MR. MANALP: First name is Y-U-K-S-E-L. Last
4 name is Manalp, M-A-N-A-L-P.
5 MS. TWOHIG: Can everyone hear? Okay.
6 MR. MANALP: Okay. Nobody is talking about the
7 car industry today, so I thought maybe I'm in the wrong
8 place.
9 MS. TWOHIG: No, you're not in the wrong place.
10 We were focussing on credit scoring in the mortgage
11 industry. But we are interested in comments that are
12 broader than that.
13 MR. MANALP: My observation has been --I'm a
14 Finance Manager of a dealership. My observation has been
15 that the credit scores provided by FICO are somewhat less
16 accurate than they would like for it to be. That's
17 obvious from the comments Caryl made there and several
18 other people. Because the basis of the information is
19 the database supplied by credit reporting agencies. And
20 if the information is wrong, of course scores are going
21 to be wrong.
22 So I have two cases that I brought with me that
23 I can just briefly describe. I had a customer who is a

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 162.
162 Page 163 164
163
1 physician. He makes a lot of money. His wife is a
2 physician. They make a lot of money. And they go back
3 to 1983 in the credit bureaus. And I counted three late
4 payments, and they still have 77 percent of their credit
5 line that is still available. There are thousands of
6 payments they made on time. And the three late payments
7 added up to $175 or something like that. And his score
8 was 585.
9 That combined with how the banks are training
10 me, or telling me what to do --they send me a piece of
11 paper at the beginning of each month. If the guy's
12 credit is such and such, this is the rate he gets. If
13 it's less than six points, they don't even look at it,
14 and it's very difficult to change their minds.
15 So I'm just going to take a couple of minutes,
16 and I'm going to read to you what one of the credit
17 reporting agencies had us sign, because we're a
18 subscriber. Okay.
19 Here we signed. The subscriber recognizes that
20 factors other than the Empirica score must be considered
21 in making a credit decision, including the credit report,
22 the individual credit application and economic factors,
23 which is what I also believed. The factors that are

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 163.
163 Page 164 165
164
1 provided by Trans Union as significantly contributing to
2 the score may be disclosed to consumers as the reasons
3 for taking adverse action, as required by Regulation B.
4 However, the score itself is proprietary, may not be used
5 as the reason for adverse action under Regulation B, and
6 accordingly shall not be disclosed to credit applicants.
7 Two things. Number one, the banks are forcing
8 us and themselves --any of us --to use it 100 percent
9 of the time. Some banks are. Number two, we're not
10 allowed to disclose to the customer, so the customer has
11 no idea what his score is, not much less try to figure
12 how to straighten it up.
13 There are errors. The example I'm going to go
14 back to --I have another example.
15 MS. TWOHIG: Would you use the mic, please?
16 MR. MANALP: Yeah.
17 MS. TWOHIG: Thank you.
18 MR. MANALP: I have another example to go with
19 it. I pulled another bureau for a student. He's a full
20 time student and has part time work. The only credit
21 line on there he had is a Cellular One for something like
22 $36 a month for ten months. He had no other credit
23 lines. He had a score of 715.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 164.
164 Page 165 166
165
1 (Laughter.)

2 MR. MANALP: The bank's idea would allow me to
3 spot a $20,000 car to this person because of his score.
4 And the bank's idea would allow me to turn down the
5 physician, unless he paid 50 percent down and he paid 22
6 percent interest.
7 The credit scores are not accurate. And I think
8 the public should be allowed to find out what their score
9 is, and that would allow them to straighten it up.
10 There's only one way.
11 MS. TWOHIG: Thank you very much for your
12 comments.
13 (Applause.)
14 MS. TWOHIG: We're running late, but I think
15 we've had some excellent speakers this morning and some
16 very important issues put on the table. The agenda calls
17 for returning from lunch break at 1: 30, which does not
18 give you much time to eat. So why don't we all try to be
19 back in an hour. I guess it's 12: 50 now, so at 1: 50
20 we'll start up again.
21 Thank you very much.
22 (Lunch recess.)
23

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 165.
165 Page 166 167
166
1
2
3
4 AFTERNOON SESSION
5 -----
6 MR. MEDINE: Thanks very much. If we can have
7 half as lively a discussion this afternoon as we had this
8 morning, I think we'll be doing very well. First, before
9 we get started with the panel, Pam Johnson, who is here
10 from Fannie Mae, will have to leave shortly, and just
11 wanted to make some brief comments based on the morning's
12 discussions.
13 Pam?
14 MS. JOHNSON: Thanks. Is this working?
15 MR. MEDINE: Yes.
16 MS. JOHNSON: Okay, good. I asked David if I
17 could just make a few comments, because unfortunately I
18 am going to have to leave, and I didn't want to leave
19 before having an opportunity to comment on a few of the
20 comments and questions that came up.
21 I guess there are two things that struck me
22 about some of the discussion this morning. One is
23 obviously there is a fairly significant degree or level

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 166.
166 Page 167 168
167
1 of suspicion and doubt about credit scoring. It goes
2 without saying. And that seems to be fueled in part by
3 personal and professional experiences, and also by a lack
4 of information. We don't know they work, and therefore,
5 you know, we're uncomfortable with that. A perfectly
6 good reason to have this forum, and I think it's an
7 excellent opportunity to bring all these issues to bear.
8 The second thing that struck me is that there is
9 a significant level of mis-information about --and I can
10 only speak for Fannie Mae and the secondary market --
11 about the role that we play, and also what our position
12 is relative to credit scoring. And perhaps I didn't do
13 as good a job this morning trying to describe our
14 position, but I would like to take another minute to do
15 that, because I think it's important.
16 Before I do that, I think I would like to say
17 that for those of you who don't know this, and I'm sure
18 most of you do, there is a difference between the
19 secondary and the primary market. Lenders and brokers
20 originate loans in the primary market. They take the
21 loan application. They do the underwriting, now of
22 course using our automated underwriting systems --you
23 know, there is a role for us there --and they sell the

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 167.
167 Page 168 169
168
1 loans to Fannie and Freddie. And we purchase loans
2 directly from lenders and seller servicers to Fannie and
3 Freddie.
4 Now, what does that mean. Well, it means that
5 as the investor, we actually are providing --we do
6 provide guidance around what types of credit quality --
7 credit standards we expect when we purchase a loan, and
8 we provide that guidance in a lot of ways. And one of
9 the ways we provide that guidance is through written
10 communications with lenders.
11 Now, many of the questions or issues that came
12 up today are directly answered in some of those written
13 communications. And I just want to bring that up,
14 because I think it's important to know what our position
15 is. For example, Caryl raised a situation with a
16 borrower who had extenuating circumstances. A very low
17 credit score, but extenuating circumstances, such as a
18 medical --some sort of medical circumstance that created
19 some difficulty with credit. Or a divorce, for example.
20 Now, if you turn to some of our lender letters -
21 -and I'll be happy to send you one, Caryl --they're
22 very specific about that being used as an extenuating
23 circumstance for making a loan. And so we try to provide

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 168.
168 Page 169 170
169
1 the kind of guidance that lenders need to know when to
2 sort of look beyond the score, and look beyond something
3 that's determined to be high risk.
4 As far as the automated underwriting is
5 concerned, the way it works is that we provide a
6 recommendation around whether or not we think the loan
7 should be approved. And for Desktop Underwriter --I
8 can't speak for Loan Prospector. But for Desktop
9 Underwriter that recommendation on average occurs about
10 75 to 80 percent of the time. So Desktop Underwriter
11 users are getting an approved recommendation out of the
12 system about 75 to 80 percent of the time, meaning they
13 do not need to further look beyond that decision and do
14 any more underwriting.
15 Now, the other 25 --20 to 25 percent of the
16 time, the message is very clear. We want experienced
17 underwriters to look at that loan, because there is
18 something about that loan, at least according to the data
19 that the system has taken in, that appears to be high
20 risk. There may be extenuating circumstances for the
21 high risk. There may be problems with the credit report.
22 And again, another specific written guidance
23 that we give back is that if the credit report is wrong

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 169.
169 Page 170 171
170
1 or inaccurate, or there is information in it that's
2 inaccurate, the lender is instructed to disregard the
3 credit score, because it cannot be right.
4 So we do provide a lot of guidance around these
5 issues. I can tell you that based on conversations with
6 Caryl and others that it does not always get translated
7 in the lending community. But I think it's important to
8 know that, you know, we are looking to buy those loans
9 that experienced underwriters and experienced mortgage
10 professionals say are good quality loans. That's the
11 business we're in, and we have every interest in getting
12 those loans from lenders.
13 On the numbers --the examples that Caryl cited,
14 you know, I have no doubt that some lenders put
15 limitations on the loans they will sell. But I will
16 state again unequivocally --did I say that right --that
17 we do not have a minimum credit score in our automated
18 underwriting system. Again, we do not have a minimum
19 credit score in our automated underwriting system.
20 So, you know, do we think that the low score
21 represents on average a high degree of credit risk? It
22 does, but that assumes that the information is correct.
23 Again, a problem we all know. And it does not

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 170.
170 Page 171 172
171
1 necessarily pick up those extenuating circumstances. And
2 we would like and hope that the mortgage professionals
3 who are dealing with those loans will work through those
4 extenuating circumstances and make those loans.
5 It behooves us to buy them as opposed to them
6 having to go to a sub-prime market, for example. We want
7 the loans that perform, and so we have the same goal in
8 mind. And I just wanted to make those comments, because
9 I know it's very easy to assume that there is --you
10 know, we're on different pages. And I know that there
11 are issues here, but I just wanted you to know we are all
12 on the same page about making more loans and as many
13 loans as possible to credit worthy borrowers.
14 MR. MEDINE: Thank you. Pam, just while you're
15 here. The FDIC before the break raised a question, and I
16 just was wondering if you had a perspective on Fannie
17 Mae's point of view. And the question was, if scoring is
18 injecting more fairness in the lending process, why is it
19 that over the years we're seeing approval rates for
20 minority applicants declining rather than increasing on
21 lending?
22 MS. JOHNSON: Well, it's an excellent question.
23 First of all, let me say this. There is no denying that

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 171.
171 Page 172 173
172
1 there is discrimination in the mortgage lending process,
2 like there is discrimination in many industries. And I
3 think this is something that we all have to continue to
4 work to combat.
5 And if you go back to the Boston Fed study that
6 was done in what, 1993?
7 MALE SPEAKER: 1992.
8 MS. JOHNSON: Thank you. You know, it's clear
9 that one of the primary reasons for mortgage
10 discrimination has to do with subjectivity of the loan
11 officer making the decision, so that they are actually
12 weighing into their decision or to their underwriting --
13 or to their decision to make that loan, they're weighing
14 in the fact that this person is African American or
15 Hispanic, etc.
16 And one of the things that we view automated
17 underwriting as helping in this regard is that it takes
18 that subjectivity out of the equation. Now, automated
19 underwriting has not been around for a long, long time,
20 and it doesn't represent a hundred percent of the loans
21 that we purchase, certainly. And so it is not
22 necessarily used as widespread as one might think. We
23 would like it to be used on a broader basis. But I think

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 172.
172 Page 173 174
173
1 over time it is our belief that the use of automated
2 underwriting will in fact have a positive effect on
3 mortgage lending in minority communities.
4 And lastly, I guess I would say that we all know
5 that minority communities are the fastest growing, along
6 with immigrant communities. The fastest growing
7 population in this country. And the fastest growing
8 opportunity for those of us in this business. Besides
9 the moral imperative that discrimination is wrong, there
10 is a business opportunity here that we certainly don't
11 want to ignore, which is, there are plenty of loans out
12 there that should be made that aren't being made today.
13 And we would like to get at that. I mean,
14 that's the business we're in. So ultimately I think that
15 we will see a shift in this issue. That's our hope.
16 That's why we keep working at this. And we're hoping
17 that the subjectivity will, you know, over time be
18 eliminated with the use of tools that don't allow the
19 person inputing the data to make that choice.
20 MR. MEDINE: Okay. Thank you very much for your
21 comments. We're going to turn now to the Is Scoring Fair
22 panel, where we're going to discuss --as we have been to
23 some extent throughout the day --whether the use of

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 173.
173 Page 174 175
174
1 scoring adequately assesses risks for various
2 populations, and how overrides are used and if they're
3 done in a fair way or not.
4 We've had the pleasure of hearing from Peter
5 McCorkell already, so Pete will be around for the
6 question and discussion period. But each of the other
7 panelists will make some brief comments and then we'll
8 engage in discussion among ourselves and with the
9 audience.
10 Bob Cook is the Senior Fair Lending Specialist
11 at the Federal Reserve Board, and has been in that
12 position since 1994. He serves as the in-house
13 consultant on fair lending policy and enforcement issues
14 involving bank exams, complaint investigations, merger,
15 acquisition and expansion applications and regulatory
16 comment and review.
17 Bob?
18 MR. COOK: Good afternoon. It's a pleasure to
19 be here. Is this mic turned on? Can everybody hear?
20 MR. MEDINE: People can hear.
21 MR. COOK: The Federal Reserve's responsibility
22 with respect to credit scoring has to do largely, of
23 course, with examination of institutions that are subject

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 174.
174 Page 175 176
175
1 to its regulatory review. We have recently, along with
2 the other agencies, issued a set of interagency fair
3 lending exam procedures. And within those procedures we
4 have a set of review steps that we use in credit scoring
5 systems.
6 We take the approach in the front end of the
7 exam now that we look for risk factors. And in some
8 instances, a lender that uses credit scoring is going to
9 find a risk factor relative --or several risk factors
10 relevant to the credit scoring system that will cause us
11 to treat the credit scored product as perhaps one of the
12 focal points of the exam.
13 Once we make that review, our primary focus with
14 respect to credit scoring systems is in disparate
15 treatment. Now, that doesn't mean that we look to see if
16 a lender has purchased a system from Pete here, where
17 Pete snuck in females as a factor in the scoring process.
18 We don't think he does that any more. He's promised us
19 he won't.
20 (Laughter.)
21 MR. MCCORKELL: When did I stop beating my wife.
22 MR. COOK: That's right. What we looked at, and
23 primarily from the disparate treatment side, is whether

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 175.
175 Page 176 177
176
1 or not there is a judgmental creep back into the system
2 that uses scoring, such that all of the promised benefits
3 of using scoring that you eliminate disparate treatment
4 because everybody gets treated the same, has kind of
5 turned on its ear. Because after the scoring process is
6 all done, you come in and perhaps ignore --but I think
7 the proper term is that you take into consideration what
8 Pam described as extenuating circumstances, or you
9 override what the score seems to be saying. The score
10 seems to be saying this is a score of 440. It probably
11 shouldn't be made.
12 But there are reasons, many of which we've heard
13 about today, why a lender might want to make that loan.
14 The extenuating circumstances can go from --as both Pete
15 and Peter said, the score doesn't predict how individuals
16 are each going to perform. It predicts how a given
17 segment of the population that made up the scoring
18 development process indicates that a similar group of
19 people would perform. But as to each individual, that
20 score is not going to say whether that person is going to
21 default or going into bankruptcy depending on what the
22 purpose of the score is.
23 So somebody with a very brief band of

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 176.
176 Page 177 178
177
1 delinquency that reflects a short term economic downturn
2 for that person might look the same to the scoring system
3 as a sort of habitual credit delinquent, and overrides or
4 extenuating circumstances provide a means to overcome
5 that. As a consequence, you can see that we kind of
6 favor it. We believe that overrides have a definite
7 place to play.
8 And the only area that we --well, one of the
9 principal areas that we are concerned about is in the
10 fair lending area. And that is, are the overrides being
11 used on a basis that doesn't take a prohibited basis into
12 consideration. It doesn't end up allowing what we call
13 low side overrides, where the score is low and it's a
14 decision to make the loan anyhow, or high side overrides,
15 where the score is 770 and the bank says yeah, but I
16 don't like his looks and denies the loan.
17 Our principal concern, as you can imagine, is
18 with low side overrides. As I said, we look in the exams
19 at risk factors. And one of the risk factors before we
20 select a credit score product would be, doesn't there
21 seem to be a lot of overrides. And that in itself is not
22 necessarily bad. Maybe this is a bank that pays
23 extensive attention to extenuating circumstances. But it

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 177.
177 Page 178 179
178
1 is a red flag for us to see a very high percentage of
2 overrides of an otherwise --of an initially scoring
3 system. So that's going to become a focal point of the
4 exam.
5 The use of the override, to the extent that it
6 seems to be producing disparate outcomes for, let's say,
7 male versus female borrowers, or minority versus white
8 borrowers, is something where primarily we looking for an
9 analysis. The examiners are doing an analysis of the
10 quality of assistance given to those people who they
11 decide to assist in overriding a score, or in taking into
12 consideration extenuating circumstances. Does everybody
13 get kind of the same effort by the lender in attempting
14 to find ways to explain why this person ought to be
15 granted a loan, or are those granted on a disparate
16 basis. So that's sort of a primer on how we are taking a
17 look at credit scoring for the most part.
18 There is --and I know Pete sort of glossed over
19 it a little bit this morning, but promised to get back to
20 it this afternoon, as he knew we would want him to --the
21 issue of disparate impact, where the system itself,
22 although it essentially treats everybody the same,
23 produces different outcomes because of the very nature of

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 178.
178 Page 179 180
179
1 the system as it relates to differences in the population
2 that it scores.
3 So, for instance, one of the factors might be --
4 and I'm not saying that it's yours. One of the factors
5 might be that people have --and I believe somebody
6 mentioned it this morning --finance company trade lines
7 versus bank trade lines, and that people with finance
8 company trade lines receive less of a score or fewer
9 points for that than somebody with a comparable bank
10 trade, perhaps of the same amount, the same duration and
11 otherwise similarly situated. To the extent that
12 minorities or single females, for instance, might be over
13 represented in the population that uses finance companies
14 more often than they use bank trades, you could come up
15 with a disparate impact.
16 So one of the concerns we do have is whether or
17 not disparate impact is an issue. And one of the
18 questions I guess that I'll toss out to Pete to respond
19 to when he gets his turn after this entire table
20 discussion, is to what extent do the Fair, Isaac's of the
21 world --and I suspect you can speak for how some of the
22 other systems operate, but certainly Fair, Isaac --go
23 when they evaluate various possible criteria or factors

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 179.
179 Page 180 181
180
1 to put in the system to kind of weigh the relative impact
2 that one factor may have against another. So that's
3 another issue that I think we do --we are concerned
4 about. I must say that it's not a factor that we have
5 the same concern about as the use of overrides in these
6 scoring systems.
7 That factor brings in another issue, and that is
8 --the disparate impact factor --and that is the
9 lender's liability for disparate outcomes. A lender is
10 responsible under a ECOA and Reg. B for disparate
11 outcomes. And that's kind of a neutral term. That could
12 be disparate treatment or disparate impact, but disparate
13 outcomes from a credit decision process.
14 If scoring --if they follow Pete's book and
15 don't use overrides, and they're rejecting females at a
16 rate of 1.6 times males, then you could say that there
17 seems to be a disparate impact. The lender, however,
18 while it's being held responsible under Reg. B for
19 explaining that disparate impact --indeed, a lender is
20 given the opportunity under sort of the order of proof of
21 disparate impact to offer a business justifiable reason
22 --a business justification for why those outcomes are
23 different.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 180.
180 Page 181 182
181
1 And yet it doesn't know what --it may not know
2 what factors are in the box. It may not know what
3 weights are there. And it may not have the capacity to
4 make an evaluation that regulators are going to believe.
5 You know, the necessary credibility to say that yes, you
6 provided a justification for the difference in outcome
7 between male and female borrowers.
8 So that's an issue which I want Pete to address,
9 because it's been raised by lenders fairly recently as to
10 what do they do. Some of them have asked, how do I know
11 that my program that I've bought from a developer program
12 has been validated. Another obligation that lenders have
13 is to make sure that their program is validated. In many
14 instances, a validation is done by the producer of the
15 system who sold the system to the lender. And we suspect
16 that those validations are done well and wisely. But,
17 again, the lender hasn't got the capacity to show what he
18 did to validate, except to point to Pete and say, well,
19 Pete tells me that my system was properly validated.
20 So there is this kind of disconnect, despite the
21 fact that, you know, as a general matter when you've got,
22 you know, a firm like Fair, Isaac, that is too big to
23 discriminate

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 181.
181 Page 182 183
182
1 MALE SPEAKER: It's like a bank being too big to
2 fail.
3 MR. COOK: Too big to discriminate because there
4 is so much at risk, that we expect that validations are
5 going to be done well and wisely.
6 Another Fed focus --
7 MR. MEDINE: You have about a minute.
8 MR. COOK: Okay. Another Fed focus is on what
9 Carroll Justice spoke about, and that's the use of
10 scoring to support a risk based pricing of loans. And
11 really we're talking in many ways about sub-prime lending
12 there.
13 To the extent that a FICO score can
14 differentiate risk levels, it offers lenders a more
15 precise way to quantify the difference in the cost of
16 that risk, or in the way that that risk ought to be
17 priced for as between, let's say, a B and B-and a C+ and
18 a C loan. And to that extent, scoring serves a very
19 useful purpose in the sub-prime lending area, a risk
20 based pricing area.
21 The danger that we see --one of the dangers --
22 is the extent to which a lender doesn't pay close
23 attention to the risk differential that the credit score

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 182.
182 Page 183 184
183
1 is describing and takes sort of a more pull it out of the
2 blue approach to pricing for risk, adding a little extra
3 just to be sure that they've got it right. And there is
4 a challenge in pricing sub-prime loans, and a lot of
5 lenders --A lenders --who try to get into it, learn
6 that it's a very different task.
7 At any rate, that's something we're concerned
8 about, is that the price differential between one risk
9 level and the next ought to bear some reasonable
10 relationship to the actual difference in risk. The cost
11 of that risk.
12 And finally, from a consumer concern not knowing
13 their score --we heard some talk about that today --I
14 think --I'm not so sure that I share that concern to the
15 same degree that it was expressed today, and I'll tell
16 you why.
17 To an individual --as opposed to a lender
18 trying to see what the band of scores are of this
19 customer for some such analytical purpose, to an
20 individual customer it's probably more valuable to know
21 that he's got a score that's low, that's hurting his
22 capacity to either get a loan or to get an A loan. It's
23 probably more valuable to know what the various reasons

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 183.
183 Page 184 185
184
1 are that are driving that score down, rather than the
2 score itself, because he then has valuable --he or she
3 has valuable information about whether it's something
4 they can do something about in a short term, or whether
5 it's something that's going to have to heal over time.
6 And Reg. B requires lenders to provide the --
7 MR. MEDINE: I'm going to have to ask you to
8 wrap it up.
9 MR. COOK: Okay. Provide the most important
10 reasons why --not only as to judgmental systems, but
11 also as to scored systems, what most adversely affected
12 the score.
13 So with that, I turn it back to you.
14 MR. MEDINE: Thank you. Thank you for
15 highlighting, I think, some of the challenges in this
16 area, where we on the one hand have heard scoring can be
17 used to eliminate some of the bias and judgmental
18 systems, but yet disparate treatment may be inherent in
19 the scoring system itself, and where overrides take into
20 account some of the sympathetic concerns and/ or unusual
21 situations, but again could also be used in a
22 discriminatory manner. So some of these are some of the
23 challenges in using a scoring system.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 184.
184 Page 185 186
185
1 MR. MCCORKELL: I hope you meant to say
2 disparate impact might be inherent. Disparate treatment
3 would not be.
4 MR. MEDINE: Well, hopefully it wouldn't be.
5 Thank you. Turning to Debby Goldberg, Debby is the
6 Acting Director of the Neighborhood Revitalization
7 Project of the Center for Community Change. The project
8 provides assistance to groups that are concerned with the
9 availability and affordability of credit and other
10 banking services in their communities. Debby is also
11 involved with the Center's public policy work on
12 reinvestment and fair lending issues.
13 MS. GOLDBERG: Thanks. I think it's been a very
14 interesting discussion here so far today. The earlier
15 panels really laid out a lot of the issues that are
16 challenges for people who are working in this field, or
17 for consumers who are dealing with these questions. And
18 at the risk of --I'll make an effort not to repeat what
19 people have said before, but maybe to give you a slightly
20 different perspective than some of the people who have
21 spoken before me.
22 One of the things that I see when folks talk
23 about credit scoring and automated underwriting systems

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 185.
185 Page 186 187
186
1 --and I think we see this from a community reinvestment
2 perspective in other ways when you talk about technology
3 and its use in the banking industry and its impact on
4 consumers and their access to banking services, whether
5 those are mortgages or other forms of banking services --
6 is that there is this tremendous promise that technology
7 holds. And I think people want to believe that
8 technology can work to everybody's advantage, and it's
9 clear that it can work to some people's advantage.
10 But I think what you learn if you watch how
11 these systems play out over time, is that you can't
12 assume that the use of technology, whether that's credit
13 scoring, automated underwriting or anything else, will
14 necessarily work to the advantage of everybody in our
15 country. And the folks who are particularly at risk of
16 either not being served well or actually being
17 disadvantaged tend to be low and moderate income people,
18 people of color and other folks who are protected under
19 our fair lending laws.
20 And so I think it's very important not to assume
21 that just because something can be described as being
22 objective, that it will be used in an objective way. And
23 I think our quest to eliminate all subjectivity from the

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 186.
186 Page 187 188
187
1 mortgage lending process as a way of making sure that
2 nobody gets discriminated, is in fact a quest that's
3 going to be vain. And I'm not sure that we ought to be
4 pursuing it in the way that I think some folks are.
5 I think we ought to be looking perhaps at other
6 ways of ensuring that discrimination and unfair treatment
7 doesn't creep into the system, because I don't believe
8 that you can have a mortgage lending process that is
9 completely objective, no matter how you go about it. And
10 I think that that has actually been illustrated by some
11 of the remarks, as I said, of the folks who have gone
12 before me, and perhaps most effectively by this last
13 panel, who made it clear that at every step along the way
14 there is --as you know from the Wizard of Oz, there is a
15 man behind the curtain. You may have this wonderful, you
16 know, image projected on the screen, but there is in fact
17 a person, or in some cases a series of people, who are
18 involved in the process and who have an impact on the way
19 that the process and the system works.
20 I want to just try and kind of highlight for you
21 what I think are some of the points along the way in the
22 use of credit scoring and automated underwriting systems
23 where bias, whether that's illegal or just unfair, may

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 187.
187 Page 188 189
188
1 creep into the system.
2 One of those is at the front end with the data
3 entry. And the gentleman over here talked about, you
4 know, garbage in and garbage out. And I think that's
5 something that we all have to keep in mind, that not only
6 are there problems with the credit reporting system in
7 this country. But as I tried to listen carefully to the
8 speakers who were explaining how credit scoring works, it
9 seems to me that there are elements of the system at the
10 front end that are subject to manipulation, which can be
11 a good thing or it can be a bad thing, depending on
12 whether it's done in a way that helps you or done in a
13 way that hurts you.
14 The Boston Fed study that someone referred to
15 earlier actually identified that as one of the key
16 factors in explaining the disparate levels of mortgage
17 lending to minorities and non-minorities in Boston. And
18 that was how much assistance were people given in being
19 on paper the best qualified applicant that they could.
20 And I think when you talk about this kind of a
21 system, you still have that issue that you have to
22 grabble with. How much help does someone get at the
23 front end to look as best as they can look on paper, so

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 188.
188 Page 189 190
189
1 that the data going into the system is going to serve
2 them in the best possible way, and are people given
3 different levels of treatment.
4 The second question --or the second place where
5 I think unfairness or discrimination can creep in is in
6 the scoring system itself. And I think that, you know,
7 the example that has been cited a couple times now, of
8 how the use of nontraditional sources of credit --the
9 household finances or others of the world --affect
10 someone's score and the way that they get treated by the
11 system, is a very good example of what the potential
12 problems may be.
13 What happens to someone who has mostly paid in
14 cash over the years and doesn't have much or any credit
15 record? You know, the kinds of things that we tend to
16 see more in low and moderate income communities and
17 communities of color, I am troubled are not necessarily
18 treated adequately by credit scoring systems. And I
19 don't think anyone will ever be able to satisfy me, and I
20 think a lot of other people, on the question until there
21 has been somebody who does an impartial and thorough
22 analysis of these systems, with access to all of the
23 details about the way that they work, and can say

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 189.
189 Page 190 191
190
1 definitely at least at some point in time that there is
2 no bias built into the system itself. And that kind of
3 scrubbing of the credit reporting system --or, excuse me
4 --the credit scoring system simply has not happened to
5 date.
6 And, you know, in this same light, I want to
7 make a different point about the use of nontraditional
8 sources of credit. Because I think when you talk about
9 fairness, one of the issues that low income people in
10 particular, and again people of color, have faced is that
11 their access to credit sources in many cases has been
12 limited, and people have been forced to use high cost
13 sources of credit, to pay more for basic credit,
14 regardless of what form that may take.
15 And particularly when you're talking about low
16 and moderate income people, although not exclusively in
17 that case, if you're paying more for credit, then the
18 chances that you're going to have problems in paying on
19 time seem to me to be increased. And I'll come back to
20 this in a minute.
21 But I think one of the issues when we talk about
22 fairness is how do we make sure that people who are good
23 credit risks get access to the best priced, best

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 190.
190 Page 191 192
191
1 structured credit that they ought to be able to have
2 access to. How do we change the dynamics of the system
3 that have led to some of the statistical results that
4 maybe we see in the analysis that's done so far.
5 And then the third point along the way where I
6 think bias and unfairness can creep in is at the back
7 end. And again, people have talked about how that can
8 play out. Once you get a score, or once you get the
9 result of the automated underwriting system, then what
10 happens. Are you --you know, has your file gone back
11 through and underwritten manually? You know, are
12 compensating factors then taken into account? Does that
13 happen on a fair and equal basis or not?
14 In this area, one of the points I want to make
15 sure we're all clear about is that there is a very large
16 part of the mortgage market that is not subject to review
17 by my colleagues who are here on this panel who work for
18 the government --the banking and regulatory agencies.
19 And so even though some lenders have bank examiners who
20 come in and look over their shoulders to try and make
21 sure that these kinds of judgmental aspects of their
22 mortgage lending process doesn't play out in a way that
23 may be biased, there are a lot of other lenders out there

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 191.
191 Page 192 193
192
1 who are never subject to that kind of review. And I
2 think when we look at the impact of credit scoring and
3 automated underwriting, we have to look at the whole
4 market and not just the regulated part of the market.
5 And then the other thing I wanted to mention at
6 this point in terms of the back end is, once you get a
7 score --once you get an outcome, how do you then know
8 that an applicant gets the best possible product that
9 they qualify for. Or to what extent, as, you know, Bob
10 Cook was just saying, does, you know, a little padding
11 get added on to the cost of that loan, whether that's in
12 the form of the interest rate or in the form of other
13 fees that may get tacked on. And who are the people who
14 get the best deals and who are the people who get the
15 worst deals. And how does somebody know what is the best
16 deal that they ought to qualify for.
17 I think the mystic that we see associated with
18 credit scoring systems is particularly problematic in
19 that way, because people don't know what they ought to be
20 able to get. And it's very hard for a consumer to be an
21 effective advocate on their own behalf, given that kind
22 of mysterious nature of the system.
23 I want to just close with two quick points. And

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 192.
192 Page 193 194
193
1 that is --and I think Margot will probably talk about
2 this more a little bit later on. But for me it's really
3 impossible to look at credit scoring and automated
4 underwriting without also then making the linkage to risk
5 based pricing and sub-prime lending. Because in the
6 communities where the groups that the Center for
7 Community Change works, low and moderate income and
8 minority communities across the country, that is a very
9 big issue.
10 The presence of sub-prime lenders and predatory
11 lenders, and the way that they use credit scoring, is a
12 very big problem that groups are really struggling with.
13 And on the one hand we have community groups who have
14 been working for years to develop the kind of mortgage
15 counselling programs that some of the earlier panelists
16 talked about --I guess Marcia Griffin, in particular --
17 as a way of taking people who in the kind of, you know,
18 system that we're talking about, might not qualify for A
19 credit and making them qualify for A credit, and getting
20 them access to A credit.
21 And I think that's a good thing. And I'm very
22 concerned that the widespread use of automated systems
23 and the fear that many --either the fear that some

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 193.
193 Page 194 195
194
1 lenders will have that if they use any kind of judgmental
2 --if they have any kind of judgmental part of the
3 process, that they're going to be, you know, straying
4 into areas of discrimination, so that they'll just kind
5 of shut that down altogether.
6 And they will be --despite all of the letters
7 that Fannie Mae and Freddie Mac may send out, they're
8 just not going to want to deal with applications where
9 somebody's credit score is too low. That and the
10 increased use --or increased activity by sub-prime
11 lenders in low income neighborhoods, I'm afraid may
12 really undermine the success and the continued benefits
13 that housing counselling I think has brought to many
14 communities across the country.
15 And I also worry about the notion that the best
16 way to deal with people who don't on their face --or as
17 a result of this kind of allegedly objective analysis --
18 look like good credit risks, is just to give them credit
19 at a higher price. I think that's a very dangerous
20 notion.
21 And I guess I've been in this business long
22 enough to be mindful of some of the lessons that we've
23 learned from FHA lending over the years, where there were

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 194.
194 Page 195 196
195
1 certain neighborhoods and certain types of people that
2 got labelled --I'm kind of simplifying the process here.
3 But got labelled as FHA communities. And the only kind
4 of credit that was available in those communities was
5 credit from FHA, which in many cases was very poorly
6 underwritten with disastrous consequences for individual
7 homeowners and for the communities in which they lived.
8 And I think if we replace that kind of system
9 with one where the only credit that is available in
10 certain kinds of neighborhoods, or just certain kinds of
11 borrowers, is high priced credit, and in some cases
12 predatory credit, that we run the risk of having exactly
13 the same problems that we suffered with poor use of the
14 FHA program in the past. And I don't think that that
15 works to the benefit of the country as a whole, let alone
16 to the individual consumers who are affected.
17 And let me stop there.
18 MR. MEDINE: All right. Thank you very much,
19 Debby. If I could follow up on one of the points you
20 made earlier, as I understand it, you're suggesting that
21 there might have been a bias in lending that occurred in
22 the past --
23 MS. GOLDBERG: Right.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 195.
195 Page 196 197
196
1 MR. MEDINE: --that might have shifted certain
2 consumers from, say, banks and traditional financial
3 institutions to finance companies at higher rates, and
4 that those consumers, as a result of that bias, were
5 ending up paying higher rates and then more likely to
6 default, and therefore a scoring system would essentially
7 perpetuate --
8 MS. GOLDBERG: Right.
9 MR. MEDINE: --some of that discrimination that
10 occurred in the past. If that's your point --
11 MS. GOLDBERG: Yes.
12 MR. MEDINE: --the question is, how do you
13 today address that concern in evaluating an individual's
14 credit worthiness?
15 MS. GOLDBERG: Well, I think, you know, one of
16 the ways to get at that is through the kinds of
17 counselling programs that were, you know, described and
18 that I talked a little bit about, where you work with
19 someone to both overcome the problems that they may have
20 experienced because they've been forced to use high cost
21 credit, so that they repair their credit record, and then
22 also to make sure that they are really prepared to become
23 a homeowner, if we're talking about mortgage credit.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 196.
196 Page 197 198
197
1 That they know what to expect in terms of future costs
2 associated with home ownership.
3 You know, one of my colleagues has called it the
4 Hechinger's tax. You know that every month you're going
5 to have to go down and spend 25 bucks at Hechinger's or
6 whatever to repair something around your house. And
7 there is not a landlord that you can call to do that, so
8 you've got to built that cost into your future household
9 budget.
10 So that people do become good credit risks, and
11 then you give them access and a shot at a loan that's at
12 the same price as, you know, somebody who has had
13 different advantages in the past --may have had. And I
14 think the experience with the programs that have done
15 that has for the most part been excellent.
16 When I talk to groups around the country that
17 are involved in housing counselling programs, and they
18 talk about the default and delinquency rates associated
19 with the portfolios of loans, you know, where they've
20 done counselling, those default rates are often as low as
21 or lower than the rest of the customers for those
22 particular institutions.
23 And so I think the success is very good, and we

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 197.
197 Page 198 199
198
1 don't want to prevent that from happening in the future.
2 MR. MEDINE: Great. Thank you very much. I'm
3 turning now to Russ Bailey. Russ is the Fair Lending
4 Team Leader in the Community and Consumer Policy Division
5 at the Office of the Comptroller of the Currency.
6 Oh, George Knight? Is George Knight here?
7 Okay. Returning to Russ. He's with the Office of the
8 Comptroller of the Currency, and he has held his position
9 as Fair Lending Team Leader since 1995. He's responsible
10 for developing and implementing the OCC's Fair Lending
11 supervision and Enforcement Program, and for providing
12 advice and counsel on fair lending issues.
13 Russ?
14 MR. BAILEY: Thank you, David. Many of you may
15 not know what the Office of the Comptroller of the
16 Currency is. It is the government agency that regulates
17 national banks and their direct subsidiaries. We are to
18 national banks what the Federal Reserve is to state
19 member banks as far as regularly examining them for,
20 among other things, fair lending matters. We appreciate
21 the invitation to speak to you today and deal with some
22 of these very thorny issues.
23 The OCC issued a bulletin in 1997. It's OCC

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 198.
198 Page 199 200
199
1 Bulletin 97-24. It deals with credit scoring issues.
2 And I put 200 of them on the table outside the door. I
3 would recommend this publication to you if you are
4 interested in credit scoring issues.
5 Let me just explain a little bit about what it
6 contains. About the first 7 and 1/ 2 pages it is fairly
7 arcane. Safety and soundness information on how to
8 develop and validate credit scoring systems and some of
9 the issues that the OCC looks at from a safety and
10 soundness standpoint. And then there are several pages
11 that discuss the OCC's concerns about some of the
12 compliance issues that arise in credit scoring. And if
13 you're more interested in those compliance issues, you
14 might want to skip the first 7 1/ 2 pages with lots of
15 detailed charts and complicated issues.
16 The basic question before this panel is, is
17 credit scoring fair. And to answer that from OCC's
18 perspective in some of the terms that have been used
19 today, I think I would have to say that it is impossible
20 to predict on an individual basis. But we think the odds
21 ratio is that credit scoring systems make an underwriting
22 system likely more fair.
23 Is that okay, Pete?

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 199.
199 Page 200 201
200
1 MR. MCCORKELL: Yes.
2 MR. BAILEY: Is that stated in credit scoring
3 terminology? We would second Bob Cook's remarks that we
4 think that the greater danger with credit scoring systems
5 is that they still can be used in a discriminatory
6 manner. It is not a panacea. Just because somebody has
7 a credit scoring system, that doesn't mean okay, the
8 underwriting system is, therefore, going to be fair.
9 We have cited four of our institutions for fair
10 lending violations involving credit scoring systems and
11 made referrals to the Department of Justice. I would
12 just like to talk about those real quickly. Debby said
13 that there always is a person behind the curtain. Well,
14 sometimes the person behind the curtain does stupid and
15 unlawful things. It doesn't happen very often, but it
16 does sometimes occur.
17 We had one institution that set up their credit
18 scoring system so that if co-applicants were married,
19 they were entered --a little j was entered in a certain
20 field in the credit scoring database. And then the
21 credit scoring system combined their credit histories,
22 eliminated redundancies, and produced one credit score.
23 It also combined their incomes and produced a combined

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 200.
200 Page 201 202
201
1 debt to income ratio.
2 But if the co-applicants were not married to
3 each other, that little j wasn't entered in that one
4 field in the system. And so what was designated as the
5 primary applicant was scored, and only the income of the
6 primary applicant was counted in computing a debt to
7 income ratio. So they set that system up to begin with
8 to treat unmarried co-applicants differently than married
9 co-applicants, and that's discrimination on the basis of
10 marital status, and that's prohibited by the Equal Credit
11 Opportunity Act.
12 We had another situation in which a bank had an
13 age-split scorecard. Now, age is the one prohibited
14 basis that Regulation B permits a creditor to take into
15 account in a credit scoring system that is validated. Or
16 in the language of the reg, empirically derived,
17 demonstrably and statistically sound.
18 But this credit scoring system had not been
19 validated. Therefore, it's use of age in an age-split
20 scorecard was inherently in violation of Regulation B and
21 the Equal Credit Opportunity Act. So they just skipped a
22 fundamental step that they had to take care of in order
23 to use an age-split scorecard.

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 201.
201 Page 202 203
202
1 We had another situation in which the same
2 scorecard was used to process all applications for
3 credit. However, if an applicant applied on the Spanish
4 language application, it took a higher score to pass.
5 Let me say that again. If an applicant applied on a
6 Spanish language application, it took a higher score to
7 pass. In the internal memos in the institution, the
8 group of people who applied on Spanish language
9 applications were referred to as "the Hispanics."
10 Whenever you get into credit segmentation, and
11 large credit card operations can have lots of
12 segmentations, you run the risk of segmenting the
13 population into some group that is highly correlated with
14 a prohibited basis.
15 Now, I've heard this --an anecdotal story when
16 I was making a presentation. Someone said to me, you
17 know, several years ago I ran across a situation where a
18 credit card operation had one scorecard. It was an
19 affinity card for subscribers to Town & Country magazine,
20 and then they had an affinity card for subscribers to
21 Ebony magazine. And they used different scorecards to
22 process those applications.
23 Now, wouldn't most of us think, whoa, there

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 202.
202 Page 203 204
203
1 might be a problem there? Okay. And I think many times
2 people don't realize how many segmentations there might
3 be in a credit operation. It's rare to find it in the
4 mortgage market, the home purchase market, that much of
5 the discussion has been about today. But it's very, very
6 common to find it in other kinds of credit products.
7 We also made a referral to the Department of
8 Justice on a case in which credit scoring was used as
9 part of the underwriting process for home improvement
10 loans. Since home improvement loans are residential real
11 estate related transactions, we had the government
12 monitoring information on those decisions, which gave us
13 the race of the applicant. And if you made a passing
14 credit score, your odds were six times greater of being
15 denied anyway if you were black than if you were white.
16 And if you made a failing credit score, your odds were
17 twice as great that you would be approved anyway if you
18 were write than if you were black.
19 We wound up actually going in and pulling a
20 sample --a random sample of files in doing our own logit
21 regression model of the underwriting decisions that were
22 made at this institution, and race turned out to be a
23 very significant variable. And the loan officers were

For The Record, Inc. Waldorf, Maryland
(301) 870-8025 203.
203 Page 204 205
204
1 simply using their override discretion to make a pattern
2 of decisions that we believe was discriminatory. And we
3 referred that case to the Department of Justice as well.
4 We have a couple of other cases in the early
5 stages. Another situation in which it looks like a bank
6 has built into their system treating married co-
7 applicants differently than unmarried co-applicants.
8 That's in the early stages, however. And we also have
9 another case in which it appears that an institution may
10 have a completely different scorecard for people who
11 apply on a Spanish language application. So we have the
12 credit segmentation problem arising again, at least
13 possibly, in a case.
14 So those are some of the kinds of things that we
15 run into from a disparate treatment standpoint. We think
16 that there is much less likelihood of a problem from a
17 disparate impact standpoint, and we believe it's a very
18 difficult issue when you talk about impact in credit
19 scoring.
20 The OCC has decided that it will not cite a bank
21 for a violation of fair lending laws using a disparate
22 impact theory applied to a credit scoring system, until
23 and unless we have developed guidance for the