Typically, they are presumed to be a result of honest customers who intended to pay back the loan in full, but for one reason or another, simply weren’t able to. 

Fraud losses, on the other hand, are immediately associated with someone who has applied for a loan under false pretenses (typically a manipulated or fabricated identity as part of a second- and third-party fraud scheme) with no intention to ever pay. 

Much attention has been paid to the rise of identity-related fraud since 2020, and it’s not hard to see why: Synthetic identity fraud (a type of third-party fraud) resulted in $20 billion in losses for U.S. financial institutions that year, and large financial institutions like PayPal have since discovered millions of illegitimate customer accounts.

At the same time, many of our customers have also had hunches about their credit losses being due to first-party fraud (where someone applies under their real identity, but falsified information like income or cashflow to gain access to a larger amount of funds). Most organizations simply don’t have the data to prove it. 

After processing millions of loan applications, our data confirms their suspicions: A significant proportion of credit loss can be traced to fraud during the application process — meaning that first-party fraud is more rampant than previously thought. 

The challenge? First-party fraud is difficult for banks, lenders, and other financial institutions to identify. With photo editing tools and other document forgery techniques, manipulating information on bank statements and pay slips is easier than ever.  

As a result, organizations are giving out loans and opening lines of credit based on falsified information. They have not been equipped to effectively detect this kind of fraud. And when the loan becomes delinquent, it’ll be written off as “credit loss.” 

In this way, first-party fraud has become a “hidden fraud loss” for many financial institutions, because these credit losses are actually a consequence of document fraud during the application process. 

Our data has uncovered one more thing: instances of first-party fraud are increasing. But before we dive too much into that, let me explain how we uncovered these hidden fraud losses. 

How first-party and third-party fraud show up in documents

A key principle we maintained when building our AI-powered fraud detection solution was explainability. Because our customers use the data we provide to make account opening and lending decisions, we’re dedicated to ensuring that they can understand it and that we can stand behind it. That’s why we show them exactly where on a document any alterations have been made.

‍

‍

But in doing so, two very different patterns have emerged.

The first falls into that second- or third-party fraud category: Names and addresses are altered on bank statements and pay stubs to obtain funds under false identities. When our customers reviewed old applications, these document manipulations would typically show up whenever no repayments were made (and were attributed to “fraud loss.”)

This aligned with our customers’ expectations; of course adopting a solution like Inscribe would help them detect more fraud and prevent losses. But the second pattern surprised them.

This matched the profile of first-party fraud. In these cases, the identity was unchanged, but indicators of financial health were tampered with. On bank statements, this could be a closing balance inflated by a few thousand dollars, some NSF fees removed, or even a loan repayment transaction altered to something more innocuous. On pay stubs, salaries were being inflated (but everything else left unchanged). These applications had passed the identity verification checks, but ultimately the loan was defaulted (and marked as “credit loss.”)

Interestingly, some loans that were revealed to have tampered documents associated with them ended up being paid in full. But in the vast majority of cases, this type of tampering led to loans that eventually became delinquent.

Our customers suddenly had the ability to uncover this hidden fraud loss, meaning that they could significantly reduce their credit losses without enforcing higher salary requirements or being more stringent with business revenue metrics. And with the data we’ve uncovered about the frequency of this type of tampering, this capability has become more critical than ever. 

What tracking document tampering patterns at scale reveals

Inscribe’s document automation and fraud detection technology is what enabled us to validate our customers’ intuition with data-driven intelligence. 

Given any bank statement or pay stub — whether digital, scanned, or photographed — Inscribe automatically extracts every important detail. This allows our customers to fully automate all documents identified as trustworthy by Inscribe, in addition to acting as a key input for our fraud detection.

Inscribe’s technology is then able to identify where each piece of information is on a bank statement (name and address of the document owner, balance information, document dates, full transaction history for the statement period, etc.). 

Similarly with pay stubs: Without any human input, we can say with a very high level of confidence whether the edits made to a document only relate to the pay information, or if the identity information has also been also tampered with.

So we’re able to automatically separate out alterations made to documents that match the pattern of second- or third-party fraud (which results in “fraud loss”) from those that match the pattern of first-party fraud (typically categorized as “credit loss”).

Applying these methods to the millions of applications Inscribe processes every month, we’ve been able to uncover some startling trends:

Of fraudulent applications made by SMBs, 50% match the pattern of first-party fraud rather than third-party fraud. This means SMB lending businesses are approving customers who may attempt repayment, but have a significantly elevated risk of delinquency.

Of fraudulent applications made for personal loans, over 30% match the pattern of first-party fraud rather than third-party fraud. These individuals are inflating their salaries, and hiding evidence of bad spending habits. Again, they present a much higher risk of delinquency.

These numbers are growing. Over the last six months, we’ve seen a 14% increase in first-party fraud.

How to detect first-part fraud and reduce credit losses 

The first step in mitigating these fraud losses masquerading as credit loss is not forcing additional income requirements onto your applicants—but rather confirming that your existing requirements are actually being met today.

That means ensuring you can trust the documentation provided during the application process.

With Inscribe’s fraud detection, this doesn’t require any additional overhead. By integrating with our API, you can automatically validate all documents you receive in seconds.

Inscribe also enables you to speed up your existing processes. Rather than manually reviewing pay stubs and other documents, Inscribe’s almost instantly extracts and returns key data points — meaning you can process a document almost instantly. 

Finally, once you’ve safely extracted key data points from pay stubs and bank statements, you can use Inscribe’s Credit Analysis^TM to transform those data points into valuable credit insights. By analyzing which types of transactions were being removed from tampered statements, we’ve been able to identify the key categories linked to credit risk and build models that automatically flag those transactions.

We may have started with best-in-class fraud detection, but this is a prime example of why we continually innovate on our solution. Inscribe’s ever-evolving solution allows fintechs to make more data-driven loan decisions more efficiently — and in a matter of seconds. Want to learn more? Reach out to speak with an expert from our team.