Banking Start-Ups Adopt New Tools for Lending

"Affirm asks borrowers for their cellphone number, their name, date of birth and the last four digits of their Social Security number. "

So they obviously have a way of getting your complete SSN just based on the last 4 digits and your cell phone#. Scary, obviously making use of data mining and public records. All kinds of crappy customer service people at the phone company, the cable company and what not have that info on file when you call. They use it to verify it's really you they're talking to. What's to stop disgruntled employees with access to our info from getting a burner phone, giving Affirm your info and taking out the maximum loan and defaulting. Pity the consumer who has to fight them on a loan they never took out in the first place. They better triangulate the source of the cell phone text response and note where in the world it came from and match it up with your current locale at least.

Toward the middle of the article:

The statement “…rather than trying to determine why, for instance, proper capitalization may be a hint of creditworthiness.” denies the essence of science and returns the analysis to the realm of magical thinking.

In the paragraph, “The data scientists focus on finding reliable correlations in the data rather than trying to determine why, for instance, proper capitalization may be a hint of creditworthiness.” please define “reliable.”

The paragraph “It is important to maintain the discipline of not trying to explain too much,” said Max Levchin, chief executive of Affirm. Adding human assumptions, he noted, could introduce bias into the data analysis.” is circular reasoning and denies the assumptions (biases) inherent in the analytic algorithms.

Actually all lenders discriminate by age. Under the Credit CARD Act of 2009 they are required by law to discriminate against anyone under 21.

Software can be useful in identifying patterns to inform decisions but should never make those decisions. The original JP Morgan (who started what became GE & single handed settled the 1905-6 run in New York) asserted that you must see the face of those you lend to and shake their hand. Never lend to someone you do not trust. He would be sick at what JPMorgan has become under the tutelage of modern day Jay Goulds.

The closer to proper individual scoring the better able to lend efficiently. "Perfect" pricing would yield the best deal for every consumer by mitigating risk for the lender. At the end of the day, the lender makes money through repayment. BTW - as much as Elizabeth Warren would disagree, people who choose to borrow have responsibility in making the decision to leverage their lives. 1/3 of homes in the US have no mortgage. Funny thing, these people never had a problem.

As I recall, S&P said, in reference to a lawsuit, that "objectivity" of assessments was a "marketing term" and should not be taken to mean they actually have objective criteria for rating something as triple A, for example.

These new criteria for potential customers look about as "objective" as the ones S&P uses.

So using data and algorithms to accurately determine a person’s credit worthiness is ‘discrimination’? What a laugh. So a lender is not supposed to want to have their money re-paid? If more ‘discrimination’ has been used prior to 2007-8 perhaps the US financial meltdown might not have occurred and we would not still be crawling out even in 2015. Have we reached a point where being given a loan is an entitlement or civil right?

I find this both scary and reassuring. The current Credit Reporting system is a complete and total fraud, so fraught with problems that it created an entirely new industry just for consumers to deal with the massive shortcomings, which use evaluative measures that are completely irrelevant and quite frankly geared to appraise which consumers are more easily fleeced with credit products than creditworthiness. Why else would closing a credit card count hurt your credit score? There is little irony that this second industry was once promoted by singing pirates.

What is still needed is a real mechanism which addresses the how billing disputes are tracked, or not, Which they are not, and it's far too easy for vendors, and especially utilities, to misrepresent legitimate billing disputes as late or missed payment, and the consumer has no real recourse.

I personally think that any business regulated by the FCC, namely your phone, cable and internet provider NOT be allowed to report to the traditional reporting agencies because these companies have such high customer service issues and the fact that they set themselves up in such a way it is almost impossible to reach a human at what they call "Customer Service"

At lest this system evaluate someone on their actual habits and not what is reported by others, and can be manipulated.

The articles's affirmation revelation(once again) is wherever you go online and however long a time you spend there is commercially available. Making us safer and honest? Wonder if someone starts here and builds up a long term profile what they can get out of it?

Its a good model for third-world areas with poorly developed consumer credit rating systems and retail banking loan programs.

I would use these tools for household rooftop solar leasing programs in order accelerate scale-out of renewables and lower portfolio risk as a precursor to asset-based securitization.

The only way a lender can make loans is to sell the loans that have just been closed. An investor isn't going to plunk down $50 mil for a mortgage pool because some wonks say the applicants using all CAPS are a poor credit risk. Geez.

Problem is - Online many things can be faked. There can be no app which can rule out these devious characters from getting a loan. I'm afraid the old-fashioned way can separate the wheat from the chaff, better hands down.
This from a tech-savvy, nerdy geek.

From a consumer's standpoint, what am I getting by going to a company like Earnest, or Affirm to borrow money? Okay, I get convenience. I can apply through any digital device while in Starbucks.

But what else am I getting? I've given them permission to parse my data, splice it up, repackage it, and sell it to other financial institutions, as well as other data brokers, and even online advertisers.

In other words, what I am getting is data raped, which is the entire real currency of Web 2.0. The business model in Silicon Valley in the past dozen years is always the same model: make an attractive, seductive, easy-to-use, "free" website, make a lot of promises about what the website offers consumers, then data-mine every key stroke the consumer enters, and exploit that data. This is the Facebook business model.

Unless you read and understand the terms of service that Earnest, Affirm and the rest of these new digital data-bandits are offering you, you don't know what you are getting that you can't get from more traditional money lenders. They are likely offering binding arbitration for any legal problems, and that binding arbitration must take place in a state of their choosing, not where you live.

Borrower, beware. It's another Silicon Valley data skim and scam scheme.

As an analogy the banking system has in effect regulated current credit report standards to slightly better then junk bond status.

Enjoyed reading this article

Using "signals" analyzed by software is completely phony and in my view even worse than a credit score. Where are we now? Most people would agree that reducing a person to a credit score is simplistic. Truly fair banking would involve being able to meet with a loan officer in person, with that loan officer having the flexibility to use their human mind to evaluate the situation, seeing things a computer algorithm that spits out a credit score and credit report might not get. So now we are going from bad to worse based on what some geek in Silicon Valley thinks about you doing something like filling out an application in all caps. This is a bad idea.

I'm genuinely fascinated by those interested in changing the way creditworthiness is attributed in 2015 and beyond. As an early retiree who lives off of investment income with spotless credit, I am unable to qualify for a mortgage or even a line of credit because I am retired but not of "traditional retirement age" and "have no job." I find it terrifying for the banking industry that someone in my scenario, who is 99.9999% likely to pay back any loan on time and in full (which I'm fully aware they make the least amount of money on), can't get a bank to even pick up the phone whereas my *actually* unemployed teenage daughter will of course be offered a gold card the second she moves off to college. I understand that banks make their money off the people who pay consistently late (if at all), but I wonder why enterprising souls haven't figured out that there's plenty of reliable income to be made off people like me who would gladly pay a reasonable interest rate, on time, in full, every month.

How about developing new software to curtail predatory, greedy executives from exploiting people for profit?

My credit score is over 800. I learned to play the game. I have no interest in using credit cards, but if the only way to get a high credit rating is to use a credit card and then pay the credit card bill (obviously more work than just paying with a debit card), I'll do the dance.

But now the dance will get more complicated. I have to pretend I am reading terms and conditions (just leaving the window open while I am reading something else), use the proper capitalization, and God knows what. Maybe someone can offer an online course on how to game the new "advanced" system? Keeps us all busy.

Collective madness, that's what it is.

Suppose, just for the sake of argument, that these new screening devices have a disproportionate effect on racial minorities. So what? Why should that matter? It's absolutely clear (i) that they were developed without any kind of racial animus and (ii) they are rationally related to the goal of screening out bad credit risks. In a rational world, that should be more than enough. To then subject them to "civil rights" scrutiny on the basis of some kind of disproportionate impact theory is simply unwise and unjust.

I use capitals when filling out forms because it is more legible. Does that make me a good or bad credit risk according to their algorithm?

This is hilarious. I learned the consumer lending business in the 1970's when loan decisions were made by PEOPLE utilizing less complete credit reporting and standardized guidelines to gain consistency. At my bank exceptions were permitted, but needed to be individually justified (such as the earning power represented by a Harvard Business School Degree). Our loss ratios were as expected with an increase during recessions and lower charge-offs during periods of prosperity. Over a 10 year period this well meaning effort to reduce "discrimination" led all banks to move to credit scoring with no exceptions, primarily to reduce the risk of "no win" class action litigation. It has taken a long time to square this circle once again.

Bank start-ups? What could possibly go wrong?

"... By law, lenders cannot discriminate against loan applicants on the basis of race, religion, national origin, sex, marital status, age or the receipt of public assistance..."
___
Disability?

I think this is terrific. Alternative (eg peer to peer) lending is already taking off to the effect that while still in infancy, the big companys are taking notice and some even competing. The more options available, like this unproven but promising option, the more competition and the better the deals, eg terms and interest rates. As long as we have just a few credit card companys who can legally charge a gazillion percent interest and without competition, the worse the american publics financial condition will be. Lots of struggling folks are not fiscally savvy and are otherwise too easily taken in.

I think its a new and interesting approval process. If you really want to judge credit worthiness, see how long the person has maintained the same cell phone number and that will be a gauge to true bill payment history.

You don't even mention interest rates!

Other research has found additional predictor variables. For instance, first name. If your first name is Safyre, you're likely to end up in jail or on welfare.

Whoops. That'll fail the "effects test."

So will some of the other variables these people are using.

It's heartwarming to see someone has noticed that Facebook and Twitter predict nothing. LinkedIn, unverified, is unduly likely to contain exaggerations and outright lies.

If you want to write an article on lenders, you should include things like term and rate. From their website:

"Choose 3, 6, or 12 monthly payments
We designed our loans to offer you flexibility and control over your budget. Affirm offers rates from 10-30% APR based on your credit. Actual rates will be shown at checkout."

SO you don't know the rate you pay until you checkout and it could be as high as 30%? Its just another lender of last resort gouging lazy consumers or those who cannot get a credit card. Nothing new here.

from Wikipedia: "Discrimination is action that denies social participation or human rights to categories of people based on prejudice."

Where is the prejudice in the program? There's simply no discrimination here.

This is MRS. It should be obvious that someone who uses proper capitalization when writing is a display of a person who follows the rules. That would indicate that a person who borrows money has a higher potential to repay within the parameters of the loan agreement; however, it is well known among those of us who have made, serviced and collected loans, both real estate and personal unsecured, that the key to obtaining the loan in the computer model application discussed herein is that some know how to game the system and will default on the loan anyway. I believe the banks are behind this lending model and that it is noble to lend based solely on one's application without a face-to-face encounter with a borrower. I also believe that the paperless, faceless application will also reveal a flaw in that the computer model will be filtering data for reports at some time which indicate that lending will be curtailed in certain locations due to the income level in that location--a form of redlining by data collection. If this lending model produces data which indicates that the very people who were openly discriminated against in lending in years past, or if a home loan were not granted simply due to the address (location) of the property, then the goal of lending to anyone without collecting the social data that produces discrimination will not be met, and can further be construed as a way of attempting to circumvent anti-discrimination laws as they presently exist.

Don't overfit the models, boys! If you include every variable that you think to be predictive, you will lose sight of the actual relationship between the variables and the outcome. Also, it sounds like you are using some variables that are flimsy. The correct use of capitals? My generation - yes. People raised on twitter? Probably not.

Best way to determine credit-worthiness? Small loans to new borrowers. See how they perform. Gradually increase credit limits, although continue to check employment status, income, and overall debt picture. Old-fashioned but it will also teach people how to handle credit responsibly. A $20,000 car loan should NOT be someone's first loan! Give them low-limit credit cards so they can learn to use credit and establish a good credit record - perhaps when they register to vote?

Can we please tell congress to regulate banks!

This is the consumer side answer to non-lending by banks. The commercial side is a now well established sector called BDCs - business development companies which came into existence because banks refuse to lend to small and midsized firms. Banks invest, they no longer lend. If you look at who has invested in BDCs, all the top holders are banks. They would rather collect the dividends than actually get their hands dirty lending money.
So now we complete the distorted picture with consumer-side non banks.

"Better risk analysis, they say, will broaden the lending market and reduce the cost of borrowing."

How wonderful! Bankers are leading us to a happier time when more ordinary people --and perhaps even an extraordinary poor person or two-- will be able to obtain loans to buy a house or a car, start a business or pay an exorbitant medical bill.

And it will come to pass because our wondrous technology is providing new ways to surveil our citizens and interpret their every move online. And those who design and use these beneficent systems will be kind and socially-conscious bankers who will have our best interests at heart.

It will undoubtedly lead to a revolution in lending in which those who need money and have decent credit will be able to get it as the banks open their hands and let the cash flow out at zero percent interest.

Unless of course the loan applicants write in capital letters or spend too much or too little time reading loan agreements or use a semi-colon improperly.

The ability to study and watch American citizens by computer and judge and assess them by secret methods is the latest ominous sign that the right to privacy is being destroyed in the land of the free.

What could be more sinister than being assessed, interpreted and judged by shadowy human agents or computer algorithms who have power over our lives and futures?

What's to stop similar technology being used to determine who can go to college, get a job, obtain insurance or rent a home?

Interesting company and I did look at their "long" privacy statement and again it's one of those written by lawyers for lawyers:) I look at them all the time and it states they do not sell your "personal" data, but what about anything they mine out there in the wild? We don't know do we? Time to license and index all who sell or share data so we know. That was the question I was left with since we have so much repackaging of our data going on. You can read my campaign on this if you like and I started it 3 years ago so we know they do or they do not sell or share and what data.

http://www.youcaring.com/other/help-preserve-our-privacy-/258776

This article is unnecessarily gushing. It describes a few new companies that have recently started lending to risky credits on an unsecured basis. The author accepts on faith the claim that the startups have developed new ways to use data to achieve better loan outcomes without offering any real specifics and despite the obvious fact that the firms only have a few months of performance data.