Nearly half the U.S. population is unserved or underserved by conventional risk methods. 50-80 million Americans have little or no credit history and are unscorable by traditional models. In addition to these credit invisible (or no-hit) and thin-file consumers, another 30% are scored below prime.
Credit score classifications around prime are named for the prime interest rate historically used as a benchmark for lending to lower risk customers. Rating above and below the prime threshold would be reasonable if the models used assess to rate default risk worked perfectly. However, these models are far from perfect and rigid adherence to legacy methods is neither accurate nor fair.
Some estimates suggest around 30 percent of borrowers are miscategorized – prime borrowers who are really subprime and vice-versa. As you move down the risk spectrum, traditional bureau scores are not much better than a coin flip at predicting credit risk.
From a fair lending standpoint, there is some evidence to suggest that traditional credit scores do not disadvantage minority populations. Yet many studies argue that the dominant credit scoring system does have a disparate impact on people and communities of color in part because the credit histories on which scores are based reflect past biases.
Driving by Looking in the Rearview Mirror
Traditional credit bureaus are anchored in the premise that past performance is indicative of future results: if you’ve paid what you owed in the past, you’ll probably continue doing so; if you haven’t then lending to you might be riskier.
This expectation failed massively during the subprime crisis. More recently, there’s evidence of statistical noise resulting from information disparities as well as widespread errors in credit reporting. Regardless of how shortcomings in the status quo are manifest, the very nature of credit data gives rise to its various limitations.
Credit scores can only reflect the information on your credit report. The prevailing scores weight factors differently with the common theme that you’re scored on what happened yesterday, last month, and last year:
Lenders generally report monthly to credit bureaus. So payments, balances, and utilization are lagging indicators of financial health. Even up-to-the-minute information on new credit – or inquiries – is historical behavior that indicates you’re applying for new debt which, if successful, means you’re more riskier going forward.
It doesn’t take an advanced degree in data science or decades of underwriting experience to see what’s missing from this data: it’s a one-sided view of risk entirely focused on obligations. You don’t build an income statement looking at costs alone, and no balance sheet starts (or ends) with debt.
Seeing the Complete Financial Picture
For most consumers, the answer is found in their bank account. More than 93% of Americans receive their pay by direct deposit. Most of the payment methods consumers utilize to pay debts and other bills come back to their bank account. Usually it’s the same account where they receive their paycheck. And of course, account balances are the assets most likely to pay existing and new obligations.
So if you pivot from the one-sided perspective of credit history to the more holistic view enabled by bank data, then a fuller view of consumer credit risk emerges:
- Is a consumer earning reliable, underwritable income?
- How does their income compare to their recurring financial obligations and other spending?
- What obligations might be too recent to be reported to bureaus or hidden from credit reports altogether?
- Will there be enough cash to cover new (and existing) debt as it comes due?
A deeper analysis of behaviors around income, spending, and balances can reveal both favorable and unfavorable risks alike:
- Is a consumer living paycheck to paycheck like 3 in 5 Americans today?
- Are they saving money or is their financial condition deteriorating?
- Does their frequency of negative balances and overdraft charges suggest a higher risk of default?
The range of interesting and actionable insights available from bank data is much longer and richer. So it’s not surprising that the potential for cash flow underwriting to predict credit risk has gained momentum in recent years. Bank data can be valuable to both verify application data and augment traditional credit scores.
Until now, lenders at all levels of sophistication have only been able to utilize this powerful alternative data by leveraging two, three, or more external partners. Most gatekeepers – or aggregators – will categorize the transactions they deliver (income types, expense types, and so on). Some offer enriched data that can be building blocks for risk models or decision rules, but even long menus of features (or attributes) stop short of calibrating against empirical risk for a useful starting point in your underwriting.
So lenders must either build their own models (with their own data) or engage a custom modeling service to understand how this new data affects risk. In most cases, you’re starting from scratch to see a consumer’s bank transactions at the time a loan is originated, then map that data to each loan’s performance over time.
NinjaEdge has created a powerful platform to not only access bank data but analyze its implications for lenders and answer the most critical questions in underwriting decisions. With a single integration, our customers can retrieve transaction data from nearly every bank account in the U.S. along with an intuitive three-digit score and underlying insights – all curated for relevance and calibrated on billions of transactions married with millions of credit applications alongside performance of hundreds of millions of dollars in loans.
If you’re not already familiar with NinjaEdge, let’s start the conversation today on how cash flow underwriting can improve risk assessment in your business to drive higher conversions and fewer defaults.