Banking & Payments
What is psychometrics, and how are banks using it?
Startups are combining artificial intelligence with behavioral finance to test borrower psychology.
“Out of your vulnerabilities will come your strength,” Sigmund Freud once wrote.
That sums up how a handful of technology companies are using psychological evaluations to help banks and other lenders come up with credit scores for people who lack traditional sources of data. For banks unable to get a reliable read on potential customers – such as a budding entrepreneur in an emerging market – a psychological profile may enable them to extend a life-changing loan.
But psychometrics – the art or science of measuring someone’s personality traits – is a branch of behavioral finance that should be treated with care. Used well, it can broaden a lender’s customer base, but banks should be mindful of how they use the data.
Gauging personalities was first adapted to credit scoring by a Harvard University research initiative called the Entrepreneurial Finance Lab, which was set up in 2006 to provide alternative credit scores in emerging markets. The lab was eventually merged with a fintech, and now operates as LenddoEFL.
Several companies have cropped up providing variations of this idea, including Innovative Assessments in Israel and RevolutionCredit in the U.S. (now known as Scorenomics). ConfirmU, based in Singapore, is a newcomer, having just graduated from the latest startup batch of SOSV’s Chinaccelerator in Shanghai.
Predicting risky behavior
Saul Fine, Tel Aviv-based founder and CEO of Innovative Assessments, is a psychologist who has been developing psychometric tools for two decades.
“It’s an exciting time to be in credit scoring,” he said. “We are part of the trend for open banking and alternative data sourcing. Psychology offers lenders another way to gauge risky behavior, particularly for the underbanked.”
- Read more:
- CreditTech after COVID
- India’s credit fintechs poised for explosive growth: Credit Suisse
- Will VC continue to back lending platforms?
Zaydoon Munir, Irvine, California-based founder of RevolutionCredit, said psychometrics has only recently gained credibility. “When I created the company in 2012, the VCs told me it wouldn’t work, because Big Data can solve the problems of credit scoring. But it didn’t, because it’s still based on backward-looking data. Psychology is about predicting behavior.”
But Munir also left RevolutionCredit and the psychometrics industry in 2019, over concerns about how it can be misused – more about that below.
These companies use variations on personality models developed by psychologists to assign individuals various personality traits. Most adhere to the “big five” model developed in the U.S. in the 1980s, which measures people against a spectrum of these factors:
- Extraversion (outgoing and energetic, versus solitary and reserved)
- Agreeableness (friendly or critical, compassionate or Vulcan-cold logical)
- Openness to experience (inventive or curious, versus consistent, cautious)
- Conscientiousness (organized or careless)
- Neuroticism (sensitive or nervous, versus resilient or confident)
Text to images to games
Innovative Assessment uses a questionnaire to peg a person’s aptitude and intent to repay a loan. The questions are not right/wrong binary responses but are meant to place people along spectrums, with data analytics to back it up, such as how long a person takes to answer a given question.
Fine says the questionnaire takes about three minutes to answer. It is usually presented by a bank or fintech lender when they lack traditional credit-bureau data, or if a potential customer receives a marginal score. It is primarily for remote customer applications or onboarding, instead of a face-to-face review by a loan officer.
RevolutionCredit and ConfirmU shifted from text to images and then to gaming – and all of these companies leverage algorithms to augment scores, blending their proprietary A.I. to generate borrower insights.
“The vision is to engage people who can’t read or write,” said Yatir Zaluski, ConfirmU’s Tel Aviv-based CEO. A questionnaire might ask someone how they react upon entering a room where everyone stops and stares. ConfirmU creates a game that analyzes their behavior in a virtual setting. “I prefer to make it a challenge, to be playful, but also to put the user under pressure,” Zaluski said.
Ethics questions in America
This approach is geared to emerging markets with a lot of under- or un-banked people and small businesses.
Psychometrics can thrive in these environments, in part because borrowers can’t obtain traditional credit scores and don’t mind having to undergo a test as part of the application process.
The U.S., where psychometrics first emerged as a tool for credit scoring, is different.
Munir says psychometrics is ultimately about profiling users, rather than predicting behavior. “Psychometrics is an input, not an outcome,” he said. “It’s like saying ‘You’re black,’ or ‘You’re Middle-Eastern’.”
Psychology offers lenders another way to gauge risky behaviorSaul Fine, Innovative Assessments
In the U.S. context, this becomes a problem. Regulators frown on profiling because it could lead to biases. Psychometrics is about categorizing a person – about identifying their mix of personality traits, and therefore determining their fundamental nature. This can open it to charges of racism or bias.
“Psychometrics is about innate traits – not behavior,” Munir said.
Munir had to build his games to get around this problem, but he says it’s difficult to use psychometrics in the U.S. For example, LenddoEFL stopped operating in the U.S. years ago. “The regulatory hurdle is too high, and consumers would hate it,” Munir said.
Zaluski agrees that the U.S. regulatory and business culture has made it difficult for ConfirmU to operate there. He says the market is too big to ignore, however: “We’re working on gamification techniques that will allow us to enter the U.S.”
Helping banks say yes
In the meantime, though, the rest of the world seems open to psychometrics. In many emerging markets, which often lack credit bureaus or other traditional sources of data, psychometrics can be a welcome means of giving people who are fundamentally responsible an opportunity to secure credit.
This doesn’t mean psychometrics is out to replace traditional credit scoring, nor is it a critique of credit risk models. It’s intended to be a supplement, so that lenders can say “yes” with greater confidence to more people who deserve a loan.
Psychometrics is an input, not an outcomeZaydoon Munir
“We keep it positive,” said Fine. “This is intended to boost a score.” It’s not meant to deny people credit.
Munir says that when he led RevolutionCredit, the company’s contracts with lenders required them to only use scores to include new customers – never to deny someone credit.
Ultimately, Munir says he grew skeptical about the way the technology was being applied; he sold RevolutionCredit and now pursues unrelated ventures in the U.S., but continues to advise startups engaged in behavioral finance.
Personalizing credit scores
Other startups say their approach is helping people and small businesses secure credit, and that psychometrics is contributing to financial inclusion while also helping banks generate revenues.
For example, a traditional lending operation will approve 10 percent to 30 percent of applicants. Psychometrics can improve this by about 20 percent, says Saul Fine – meaning a lender who normally approves 10 percent of loans will now approve 12 percent. It’s not a radical jump but it does increase business, which improves revenues, while providing finance to more people who deserve it, particularly when scores are combined with other forms of data.
The vision is to engage people who can’t read or writeYatir Zaluski, ConfirmU
But Zaluski says ConfirmU’s data can be used by banks to decline borrowers.
In the context of an emerging market with little else in the way of data, this may be necessary. It suggests, however, that while ethics in A.I. may be relative, they’re still real, even in markets in which psychometrics or other credit-scoring tools are necessary.
Banks that are looking to deploy psychometric credit scores should therefore be careful to use these in a positive, constructive way. While the U.S. will likely remain unique in its attitudes towards the practice, questions around ethical A.I. are growing, led by regulators in places like Singapore.
That said, psychometrics seems bound to take its place among various forms of artificial intelligence-enabled business models, particularly in emerging markets. That’s because it’s about more than just coming up with a number for a credit report.
Fine believes the practice can help lenders be more confident about embedded finance and other ways of accessing new and unknown customers. “The challenge,” he said, “is to get past the score and understand the possibilities of personalizing financial services.”