Artificial Intelligence: Boosting Your Odds for Getting a Loan?

Summary

WSJ Your Money Briefing

Episode: Artificial Intelligence: Boosting Your Odds for Getting a Loan?

Date: March 5, 2019
Host: J.R. Whalen
Guest: Anna Maria Andriotis, WSJ Reporter

---

Episode Overview

This episode explores how credit card companies, notably Discover, are increasingly utilizing artificial intelligence (AI) and alternative data sources in their personal loan business. The main focus is on how AI is changing the loan application process—potentially increasing loan access for some while tightening standards for others—by identifying fraudulent applications and uncovering new risk signals. The discussion covers both the technical aspects of new AI tools and their broader implications for borrowers, banks, and the booming personal loan industry.

---

Key Discussion Points & Insights

1. How AI is Transforming Loan Applications

• Goal: Discover will start using AI to analyze hundreds of new data characteristics for improved risk assessment.
• Impact:
  • May improve approval chances for some applicants who have been denied under previous criteria.
  • Conversely, could make it harder for others to qualify for a loan if flagged as high-risk or potentially fraudulent.

"What this will do is improve the odds for some people and lessen the odds for others...The main goal here for Discover is to identify people who pose a risk of not paying back a loan...or because they're fraudsters and to catch that before giving them a loan."
– Anna Maria Andriotis [02:29]

---

2. New Fraud Detection Methods

• Behavioral Data: AI examines applicant behaviors for inconsistencies.
  • Example: If an applicant enters their employer’s full legal name (“Inc.,” “LLC,” or “Co.”), it may signal copy-paste behavior, hinting at identity theft or fraudulent intent.
  • Phone Data: Use of Internet-based phone services (e.g., Skype) to contact customer service during application can be a red flag, as these are harder to trace than landlines or cell phones.

“If the loan applicant writes the full legal name of their employer...that would be a red flag for Discover because it could suggest the loan applicant is not truly employed by that company, but rather just maybe doing a copy and paste job as they're filling in a loan application with someone else's identity.”
– Anna Maria Andriotis [03:41]

"If...Discover sees [the applicant is] calling from an Internet based phone service...that’s also going to raise some red flags...finding those people, tracing those people back...is a lot harder than a landline or a cell phone."
– Anna Maria Andriotis [04:35]

---

3. Surprising Risk Factors: High Income as a Red Flag

• Unusually High Income Declarations: Applications listing high (six-figure) incomes are scrutinized more carefully, as they are atypical and could suggest exaggeration or fabrication.
• Behavioral Analysis: Lenders now supplement traditional credit reports/scores with alternative data to gauge borrower honesty and likelihood to repay.

"Six figure incomes...are not representative of the majority of US population...when applicants put in what appears to be a relatively high income...that sends up red flags...maybe they're just putting in a high figure so that we can think of them as...a great credit risk."
– Anna Maria Andriotis [06:12]

---

4. The Growing Challenge for Lenders: Fraud in Personal Loan Markets

• Industry-wide Problem: Personal loan fraud (especially “synthetic” or fabricated identities) has increased with industry growth, making more sophisticated risk assessment essential.
• Discover’s Business Motivation: AI is expected to help reduce losses, especially since personal loans have become Discover’s worst-performing product, even outpacing credit card and private student loan losses.

"There's a whole bunch of competition for borrowers...a lot of fraudsters are getting in the mix here in terms of people...applying for loans with made up fake identities...that's where things like putting in a full legal name of a company could raise a red flag..."
– Anna Maria Andriotis [08:10]

• Data Point: In Q4 2018, Discover's personal loan losses rose by more than 80 basis points year-over-year.

---

5. Looking Ahead: Industry Implications

• AI-driven alternative data and behavioral analytics are now indispensable in managing ballooning consumer loan portfolios.
• These new risk tools will likely spread across the industry as lenders seek to balance growth, access, and risk.

---

Notable Quotes & Memorable Moments

• "Who would have thought that there'd be a benefit from landlines in the year 2019...if you're calling from a landline or a cell phone, that could actually help you from an applicant standpoint..."
  – Anna Maria Andriotis [05:22]

• "What is really different here...for decades, large lenders...primarily relied on credit reports and credit scores...now...are moving towards using so called alternative data..."
  – Anna Maria Andriotis [07:26]

---

Timestamps for Important Segments

• [02:29] – Anna Maria Andriotis explains AI’s dual impact on borrower risk and approval odds.
• [03:41] – Detection of fraud via employer names and applicant behavior.
• [04:35] – Landlines and the use of phone data as additional fraud indicators.
• [06:12] – High-income claims raising suspicion for potential fraud.
• [08:10] – Broader context: Increasing fraud and losses in personal loan markets.
• [09:45] – Summary of Discover’s incentive and industry-wide trends.

---

Episode Tone

The tone is practical, investigative, and slightly cautionary—highlighting both the promise and pitfalls of AI in lending. Anna Maria Andriotis offers clear explanations with real-world examples, while J.R. Whalen asks concise, clarifying questions to dig into consumer impacts.

---

This summary encapsulates all key points for listeners interested in how financial technology is reshaping personal lending, both for applicants and for banks battling rising fraud.

Transcript

[Small Business Owner]

Access to affordable credit helps me pay my employees, but I don't really need it.

[Retail Industry Representative]

Inflation is killing me, but who cares? Big retailers are making record profits. That's why we support the Durbin Marshall credit card bill.

[Small Business Owner]

See banks and credit unions help small businesses make payroll. This bill would cut the vital resources

[Retail Industry Representative]

they need while increasing megastore profits. They deserve it, don't they?

[Electronic Payments Coalition Spokesperson]

Tell Congress stop the Durbin Marshall money grab for corporate megastores paid for by the Electronic Payments Coalition.

[J.R. Whalen]

With your Money briefing. I'm J.R. whalen at the Wall Street Journal in New York. Credit card companies are turning to artificial intelligence to weed out the bad guys trying to fraudulently apply for loans. We'll have details and explain how it would affect your application for a loan in a moment. First, these money in market stories you should know. A study by the online mortgage marketplace LendingTree indicates the state that someone lives in is a significant factor in the mortgage consumers wind up with. And it comes down to local competition and business costs. Currently, the national average for a 30 year fixed rate loan is 4.84%, but California has the lowest average mortgage rates in the nation at 4.74%. The next best performers, rate wise, include New Jersey, Washington State and in Massachusetts. New York ranks worst on the list where rates average 4.96%. And the Empire State is followed at the bottom of the list by Iowa, Arkansas and Oklahoma. And check out the Wall Street Journal's tax calculator just in time for the new tax law to affect your tax return. Type in data like marital status, income deductions and medical expenses and the calculator will tell you how much your tax liability has changed confirmation compared to the previous tax code. It'll also tell you how much your taxes are likely to change when the tax code expires in 2027. Check it out on WSJ.com. Artificial intelligence technology is coming to the loan business. The people behind Discover Credit Cards will be using the new source of data to improve its personal loan business. And Wall Street Journal reporter Anna Maria Andreotis is on the line with us with some so Anna Maria, do you think this might improve the odds for consumers trying to get a loan?

[Anna Maria Andreotis]

What this will do is improve the odds for some people and lessen the odds for others. Discover is going to be looking at hundreds of new data characteristics about people and it's going to be doing that to primarily identify loan applicants who who present risk and essentially would increase the company's chances of people not paying back the loans that they could be Given the strategy could also help some people who would not qualify for the loans with the original data that the company was using to get approved for them. But ultimately, the main goal here for Discover is to identify people who pose a risk of not paying back a loan because they really can't afford it or because they're fraudsters and to catch that before giving them a loan.

[J.R. Whalen]

Yeah, it's interesting, you're right in your story that they're hoping that this AI technology could weed out people using putting in fraudulent information or false identity and really connecting those dots and taking them out of the running.

[Anna Maria Andreotis]

If the loan applicant writes the full legal name of their employer on their loan application. So a name that let's say ends in Inc. Or LLC or co, whatever it might be, that would be a red flag for Discover because it could suggest that the loan applicant is not truly employed by that company, but rather just maybe doing a copy and paste job as they're filling in a loan application with someone else's identity. Because when you think about it, a lot of this is behavioral. If you're filling out a loan application and you're asked for your employer, you are probably going to use the everyday name that your employer goes by and is known. You're not going. The chances of you putting in the full legal name aren't so high.

[J.R. Whalen]

Another example, I see what you're saying.

[Anna Maria Andreotis]

What they're going to be doing here is to spot signs of fraud. If the loan applicant calls Discovery customer service and is getting the application started and Discover sees that they are calling from an Internet based phone service. So not a landline, not a cell phone, something along the lines of Skype or other types of services that are Internet based, that's also going to raise some red flags. Because finding those people, tracing those people back to a phone number when they're calling from an Internet based phone. Phone service is a lot harder than a landline or a cell phone.

[J.R. Whalen]

So the good old fashioned landline actually could live to see another day here.

[Anna Maria Andreotis]

That's interesting that you say that, because that's true. I mean, not to get into the phone sector, but who would have thought that there'd be a benefit from landlines in the year 2019, as many people are really not just cutting that cord and living with their cell phones. But yeah, here if you're calling from a landline or a cell phone, that could actually help you from an applicant standpoint because it will give Discover more confidence that you are who you say you are, they can trace you back to actually your own individual identity.

[J.R. Whalen]

You also mentioned your story that loan applicants with higher incomes than others might be seen as more risky. The conventional wisdom would be that if you have a higher income, you might have a better way, a better chance of paying the loan back. So why is that?

[Anna Maria Andreotis]

It's primarily because six figure incomes, while they do sound, as you said, as very financially stable, are not representative of the majority of US population. Meaning that the low six figure incomes or higher are significantly higher income than the median U.S. household income. So apparently when applicants put in what appears to be a relatively high income, at least compared to most of the US population, that sends up red flags and it's well, is this loan applicant really making this level of money that they claim to be making or are they just putting in a high figure so that we can think of them as, oh, they must be a great credit risk because look at how much money they make. Of course they'll be able to pay back this loan. So a lot of this stuff comes down to a behavioral analysis and that's what is really different here. I mean, for decades, large lenders the size of Discover and most US banks, they primarily relied on credit reports and credit scores to determine whether to approve people for loans. Now these companies are still using all that data, but increasingly a number of large lenders are moving towards using so called alternative data, data that can give you more insights into the likelihood that somebody is going to repay the loan that they're applying for.

[J.R. Whalen]

So from a business standpoint, Discover is best known for its credit card business and personal loans, as you mentioned in your story, are among the worst performing of its credit offering. So they're really hoping the AI data can help them pare back losses that they're suffering and improve that part of the business.

[Anna Maria Andreotis]

This effort is primarily a way for Discover to root out risk to find applicants who maybe it would approve with the underwriting standards it has had all along and to use this extra data that would help it determine, well, wait a minute, somebody who looks pretty credit worthy. Actually when we look at this additional data, there's some red flags here. Maybe we shouldn't approve them. And really the problems that discovered is dealing with in the personal loan world are pretty emblematic of what's going on in the personal loan industry. This is a booming market. Outstanding balances industry wide are at record levels. In recent years there have been a number of banks that have jumped into the space to compete against fintechs that have been in the personal loan market Starting shortly after the last Financial downturn. And so there's a whole bunch of competition for borrowers and to keep originating loan volume. And what's happening, quite frankly is that a lot of fraudsters are getting in the mix here in terms of people who are applying for loans with made up fake identities. This is something that Discover see CEO spoke with me about as I was working on this article, that one of the things that they're trying to get a handle on is these so called synthetic or made up identities. So that's where things like putting in a full legal name of a company could raise a red flag because, well, wait a minute, would a real person actually do that? So personal loan losses at Discover and industry wide happen on the rise. And as you said, Discover's personal loan losses are the, the worst loan performer at the company, period. They're performing worse than credit cards, worse than private student loans that Discover offers. And just to give you an idea of what that means, loan losses on Discover's personal loans in the fourth quarter of 2018 rose by more than 80 basis points. That's a lot on a year over year basis. So clearly what's happening here is the company's trying to take additional steps to get that under control. And it remains to be seen how helpful the strategy will be.

[J.R. Whalen]

All right, that's Wall Street Journal reporter Anamaria Andreotis on the line with us. Anamaria, thanks for coming on the show.

[Anna Maria Andreotis]

Thank you.

[J.R. Whalen]

And that's your money briefing. I'm JR Whalen in New York for the Wall Street Journal.

[Small Business Owner]

Access to affordable credit helps me pay my employees, but I don't really need it.

[Retail Industry Representative]

Inflation is killing me, but who cares? Big retailers are making record profits. That's why we support the Durbin Marshall credit card bill.

[Small Business Owner]

See, banks and credit unions help small businesses make payroll. This bill would cut the vital resources

[Retail Industry Representative]

they need while increasing megastore profits. They deserve it, don't they?

[Electronic Payments Coalition Spokesperson]

Tell Congress stop the Durbin Marshall money grab for corporate megastores paid for by the Electronic Payments Coalition.