Alternative Data Has Proven Effective as Next-Generation Data for Mainstream Use in Credit

Share |

More than 50 million Americans don’t receive credit scores because of insufficient or nonexistent credit information. Consider what this means – nearly 15 percent of the adult population can’t get a mortgage, rent a car or use a credit card. But this barrier, which limits personal economic power and holds back our economy, is increasingly being broken-down by the use of new, alternative data sources and modeling techniques.


This is not a sudden or radical shift, but one that has been happening for quite some time.  At this point, the evidence demonstrates that data sources currently residing outside of the traditional credit bureau files are just as effective at qualifying more consumers for mainstream credit.  Therefore, SIIA’s often refers to this data as “Next Generation Data.”


New data and modeling have gained enough attention that regulators are seeking to learn more about opportunities and outcomes.  Today, SIIA submitted comments to the Consumer Financial Protection Bureau (CFPB) in response to their inquiry about the use of so-called “alternative data.” The CFPB is right to be assessing the current and future market developments, including benefits and risks, and how these developments could alter the marketplace and the consumer experience.


SIIA’s speak heavily to the benefits that have transformed “alternative data” to “next generation data,” with a clear result of higher credit approval rates and lower defaults.  That is, financial institutions employ alternative data to gain additional insight, or to perform what has been described as a “secondary look” at consumers who were initially declined credit either because of poor credit history or a lack of credit history. In this scenario, the lender may find that while a traditional credit score is low or non-existent for a given consumer, the Alternative Data score may be high due to the inclusion of previously overlooked variables. The lender can now successfully extend credit to the consumer while maintaining appropriate risk levels in their portfolio.


According to research from McKinsey & Company, “new alternative data models have cut credit losses in experimental forays into lower-income segments by 20 to 50 percent and doubled their application approval rates.” Another study on an auto lending portfolio found a model that uses only credit bureau data would lead to approval of 74% of applicants at a default rate of 3%. By incorporating alternative Data, the lender could increase their approval rate to 85% while still maintaining a 3% default rate resulted in up to 24% more loan approvals than a bureau only model.  Clearly, alternative data provides a substantial benefit for both consumers and lenders.


Additionally, research by FICO found that with the majority of the consumers who have scant or stale bureau data, traditional scoring would not provide for them to establish credit.  About 65% of these consumers have a negative item and no active account.  With no positive data flowing into their files to offset the negative, they would likely score too low to obtain credit.   Additionally, the same research revealed two key elements about consumers that become scorable as a result of alternative data.  First, more than a third of these individuals receive a score of 620 or above. Second, and perhaps more importantly for the long-term credit opportunities of thin-file individuals, the majority of these applicants granted credit go on to manage their credit obligations responsibly.


The challenges with scoring no-file or thin-file individuals is that it differentially affects historically disadvantaged minorities.  A recent Lexis-Nexis study found that 41% of historically underserved minority populations of Hispanics and African-Americans could not be scored using traditional methods, while the unscorable rate for the general population was only 24%.  Minorities face an unscorable rate that is 1.7 times the rate – almost twice – the rate for the general population.


Alternative data sources can substantially help to provide accurate scores for many of these individuals.  For instance, an alternative credit score, called RiskView, built by Lexis-Nexis relies on public and institutional data such as educational history and professional licensing, property asset and ownership data such as home ownership, and court-sourced items such as foreclosures, evictions, bankruptcies, and tax liens.  The Lexis-Nexis report demonstrated the extent to which credit risk scores built from alternative data can help to extend credit to unscorable consumers, finding that fully 81% of the unscorable minorities received a RiskView score.


Moreover, incorporating alternative data sources into traditional measures of calculating credit scores can differentially increase access to credit for low-income and minority consumers without over-extending credit. For example, with the addition of utility payment data in VantageScore’s credit score, acceptance rates to access credit for those making less than $20,000 per year, African-Americans and Hispanics increased by more than 20%. Notably, incorporating utility payments also did not “over-extend credit.” In other words, the Political Economy Research Center (PECR) found that these higher acceptance rates did not offer credit at unaffordable prices or to those who could not afford to repay their loan.  Clearly, a major benefit of alternative credit scores is the improvement in the availability of credit for historically underserved minority groups.


Fortunately for consumers, regulators are already equipped with adequate consumer protection tools for the use of alternative data and modeling.  Specifically, due to the broad nature of the Fair Credit Reporting Act (FCRA) and the Equal Opportunity to Credit Act (ECOA) companies already have legal and regulatory incentive to evaluate non-traditional data and modeling techniques for fair lending risk, just as they do with traditional data and modeling techniques, including disproportionate adverse impact on a protected class such as those defined by age, sex, race, and ethnicity.  The FCRA requires credit bureaus (and other businesses that report information on consumers for employment, credit, or insurance purposes) to maintain standards designed to ensure “maximum possible accuracy” and completeness with respect to consumer information that they report.  Covered businesses must provide consumers with access to the data and have a process in place for investigating and resolving consumer disputes in a timely manner—this infrastructure needs to provide for compliance at a significant rate when used in high-volume scoring.  Businesses’ obligations with respect to non-traditional data are the same as those which apply to any other kind of data currently provided in a consumer report, including those relating to disclosure of information to consumers, accuracy or completeness of information, and consumer rights of access and correction. 


ECOA prevents discrimination in the granting of credit, and it creates liability for both intentional (e.g., race-based) and unintentional (e.g., disparate impact) theories.  Credit systems must be ‘‘empirically derived,” as well as “demonstrably and statistically sound’’ in order to be used without fear of liability.  A careless approach in the use of such data can easily lead to lawsuits.


As we move forward in the credit space, alternative data will increasingly be recognized as next generation data.

David David LeDuc is Senior Director, Public Policy at SIIA. He focuses on e-commerce, privacy, cyber security, cloud computing, open standards, e-government and information policy. Follow the SIIA public policy team on Twitter at @SIIAPolicy.