AUTOMATED UNDERWRITING VS MANUAL UNDERWRITING

Just this month the CFPB (Consumer Financial Protection Bureau) reported FIVE Federal Agencies are looking into how Financial Institutions use Artificial Intelligence to support risk evaluations before approving a consumer request for a financial loan. Ironically, the investigation will indirectly involve some of the elements mentioned in last month’s article titled the 9 Credit Scores You Never Knew Existed by providing insight into how a consumer’s credit behavior as well as credit reporting errors support quick pros and cons within the automated decision-making process.

The early question, is how does it affect consumers? How is AI affected by a creditors input of adverse data? Could the FICO score be considered AI? Can it disrupt loan approvals when data is input incorrectly by your creditors or whom you do business with? Let us take a closer look at one example that is a hot topic right now by using a borrower identified as Sam to provide insight in how the insertion of correct data will negatively impact his credit report by using one of Sam’s creditors. Sam has 1 collection account reporting on Equifax, Experian, and TransUnion from the US Department of Education. Sam has been avoiding his owed debt for years, finally the US Department of Education closes Sam’s account because of no payments received for the last 270 days or 9 months. The account reports closed/transferred with a zero owed balance reporting a 120 day late pay within the account/pay status section and the last pay history field reports the same. Sounds like Sam is getting a FREEBIE because he is not paying his owed debts however the account then is shuffled to 1 of 8 collection servicers for the US Department of Education. The new servicer reports a new account under the same adverse section of Sam’s credit report with the last reported owed balance before it was closed/transferred. Finally, Sam decided to do the right thing, he enrolls in a Rehab program with the new assigned servicer of the US Department of Education for a standard of 9 to 10 months. During which time, Sam pays on the loan each month responsibly. The Servicer of the debt closes out Sam’s loan as promised and then places Sam’s account back with the US dept of Edu. The Dept of Edu then reports a new loan with a new account number for Sam to payback. Six months later Sam has been actively doing the right thing making it a total of 15 to 16 months of good pay history towards his original owed debt.

Sam is feeling good about his life, decides after 16 months he wants to apply for an Auto Loan. He checks his credit scores online using a credit monitoring site that he heard about on the radio. Its free must be a good score. Sam signs up online using his phone, reviews his credit scores, gets all excited because it tells him he has a 620. Best he has ever had. Sam visits his local auto dealer, they find him a car, pull his credit but unfortunately, he has a 540 AUTO FICO score. Sam does not really understand, he has been paying on his Student Loan account for the past 6 plus months plus he has a 500.00 secured limit credit card that he only owes 200.00 towards, he makes good money and can afford a car payment. The car manager does their best to explain the issues but has never been to school or formal training in evaluating credit scores and has no real answer other than he or she has seen this a thousand times. So, what can Sam do to help himself? Sam has 3 accounts. 1 open and 2 closed however the 2 closed are from the original debt Sam is paying on. That debt is now good except the account has a new account number which should have removed the first closed account as a duplicate and or removed Sam’s default history which is located in the pay or account status as well as Sam’s default pay history section, remember Sam did complete the rehab program.

All along both negative accounts are in COMMUNITY. Sam’s adverse section of his credit report keeping Sam in a Credit Prison Trap for at least 5 to 7 more years (Depending on the state you reside. N.Y. is 5 years all other states are 7 years) that will remain in a delinquent status. Effecting Sam’s credit scores. Sam created a debt he must pay for his actions. I can’t disagree with that debate however if you look at the problem deeper and then ask yourself, why 2 adverse accounts vs 1? One of those closed/transferred while reporting 120 days late in the pay status section as well as his pay history section> Would not that be considered double jeopardy? The second account reports as a closed rehabilitated collection account. Should both remain on Sam’s account for 7 years? How can you determine a Risk Evaluation without first fixing the problem? How is the data uploaded, is this correct data, why is the correction process difficult? Why are the bureaus governing themselves? These are publicly traded companies each stock valued from $90 to $1300 per share. That is impressive, especially when you compare the 3 Big Tech Firms, Apple, Facebook and Google are valued less than Equifax, Experian, TransUnion and they remain under the radar. These rules below describe why Sam and others like him should not be overlooked and in my expert opinion why he should not be denied credit when the rules clearly are not being used to benefit Sam. Per the CRRG (Consumer Resource Reporting Guidelines written by former employees of each National Repository known as Equifax, Experian and TransUnion) codes duplicate account reporting are discouraged except within the Federal Student Loan Providers. Per the CRRG codes the last known action should be reported in the pay status or account status however, the US Department of Education has its own rules. Should these rules be applied to Sam’s Credit Report so that he can objectively have his credit report examined manually?

  • Per the Department of Education Rules, it clearly specifies that the 270 days or 9 months of reported default months to be removed from the record. Department of Education Rules 20 USC § 1078–6 (C) Upon the sale or assignment of the loan, the Secretary, guaranty agency or other holder of the loan shall request any consumer reporting agency to which the Secretary, guaranty agency or holder, as applicable, reported the default of the loan, to remove the record of the default from the borrower’s credit history.
  • Per the following Regulation 34 CFR § 682.405 (b)(2) The guaranty agency must report to all national credit bureaus within 90 days of the date the loan was rehabilitated that the loan is no longer in a default status and that the default is to be removed from the borrower’s credit history
  • For Private Loans that have offered a rehab program (I) the financial institution chooses to offer a loan rehabilitation program which includes, without limitation, a requirement of the consumer to make consecutive on-time monthly payments in a number that demonstrates, in the assessment of the financial institution offering the loan rehabilitation program, a renewed ability and willingness to repay the loan.

After reviewing the above rules, it does not appear the US Dept Rules are being applied universally for Sam. In my opinion the default pay status demonstrates the same definition as the borrower’s default pay history. Sam’s Credit Report will never make it to a manual underwriter objectively because he will be viewed as a high risk due to too many collection delinquencies which places Sam at higher risk. More importantly his credit scores will stop any manual underwriting. The main reason is FICO credit scores do not understand the hot plate is hot, it only knows what has been placed within Sam’s credit files. So, when it recognizes adverse items it scores those items based on 5 known ingredients; age, pay history, balance, type of credit and new credit. All Five effecting Sam’s ability to rebound because of how his Student Loan is being reported. What Sam may learn is that FICO until June 2023 is the only authority and credit score provider utilized in 3 specific industries namely Auto, Mortgage, and Credit Card as well as engaging in new developments, improved coding, more specific algorithms to specific industries mainly approved subscribers to benefit from its educational pay to play events. So, with this new information Sam is questioning why lenders would even take the time to examine the problem when the issue is not on the lower level of management but the corporations that rely solely on quick risk evaluations to expedite the workload, increase customer relations as well as justify rate and terms provided to the consumer when/if approved. So how is the problem corrected and how can others overcome an error that does not agree with bureaucratic policies based on risk evaluations?

Then you must conclude that the FICO credit score is not manually underwritten they are used primarily in the Automated Underwriting process. Manual underwriting is usually performed when the scores have passed but the other areas of the loan require a closer exam like DTI issues, Asset Issues, Employment Issues which has no bearing on credit report data. Your Creditors who subscribe to the credit reporting agencies are responsible to report your data accurately and are required to follow industry standard codes that best describe your last known entry if they are to qualify as a subscriber to Equifax, Experian and or TransUnion. Unfortunately, Sam is denied access to an automobile loan per the Automated Rules built within the lenders guidelines to secure a positive credit score grade.

This Artificial Intelligence has been adopted in many forms however how effective can a Lenders Automated Underwriting perform when utilizing AI to predict a future outcome based solely on what data has been provided to accurately determine a consumer’s credit worthiness may be questionable.

The Automated Underwriting can be a great source of technology if its operating based on fairness and equality, but the disadvantages currently outweigh the pros when you take into consideration how it can financially cripple a large sector of our population when credit reporting errors are not corrected or worse, unknown.