How Alternative Data and Psychometrics are Transforming Student Lending in Brazil
An Interview with Carlos Furlan, CEO @Pravaler
As part of our latest interview series with global influencers in online lending and alternative data, we have been fortunate to speak with leaders who are transforming the financial services industry. Today is no exception. We are excited to have the opportunity to chat with Carlos Furlan. Carlos is the CEO of Pravaler, the largest fintech of financial solutions for education in Brazil. He has held this position for nearly a decade, and has been with the company for nearly 2 decades. During this time, Carlos has led the process of creating and developing one of the most successful and growing names in the industry.Pravaler is unique to the lending space, as it focuses on a traditionally challenging credit market – student education. The company has serviced nearly 200,000 students, and partners with over 500 institutions around Brazil. Pravaler has made a huge impact by making education more accessible, and transforming the lives of millions of students and their families.
Carlos, let’s start with the most obvious question: Students typically have little or no credit histories, and that makes it very difficult to assess their credit risks. What is the solution?
“In Brazil, historically, credit models are based on negative information from customers, that is, we tend to signal customers who have experienced some difficulty in the administration of their previous debts and thus make the decision to approve or not the credit. However, this type of analysis ends up generating doubts such as: the customer does not have negative information, because he has a good history or his history does not exist or is not available?Faced with this deadlock, we have dealt with the risk assessment requesting a cosigner in our loans, who is usually a relative of the student (especially parents) who already has a more consolidated financial information history.Another important distinction of our program is the students’ payment flow, simultaneously minimizing Pravaler’s risk and accommodating the needs of lower-income students. We finance students’ educations through successive small loans. Once a student is approved to borrow, we issue an initial loan covering part of the first semester, an then another one to cover the rest of this first semester. Each subsequent loan covers another semester’s tuition and is repayable in a period of 12 months. While each of these semester loans is independent from each other, repayment is coordinated and staggered, such that only one installment is due each month. In addition, the availability of positive registration data, which was implemented in Brazil two years ago and is now gaining traction, also helps us in the use of behavioral models, such as Worthy Credit.”
“Alternative credit data came to complement our standard credit analysis. Currently, we use alternative data in about 30% of the received proposals, and we could increase our approval in 10% with its use.”
That is very interesting. How has alternative credit data changed credit models for student lending?
“Alternative credit data came to complement our standard credit analysis. Currently, we use alternative data in about 30% of the received proposals, and we could increase our approval in 10% with its use. Thus, we have been using the psychometric model as a way to analyze credit risk with a differentiated dynamic, helping us to enable access to students who are not approved in a first analysis.Since 2020 we have been using the Worthy Credit psychometric model, and in June of this year, we implemented a new model with information from Positive Registration, both as a complement to our standard analysis. For the future, in addition to expanding Worthy Credit to other Pravaler products, we want to invest in data from Big Data sources, such as digital data, household information, relationships network, as well as immerse ourselves in the world of opportunities that Open Banking will bring in Brazil.”
“We noticed that the public with the best ranking within Worthy Credit presented a default 50% lower in comparison to the other students approved by the standard analysis, which proves the effectiveness of the model.”
What is your opinion about psychometric data in credit scoring?
“The psychometric model has been an important tool for expanding access to credit, because the score presented good discrimination on the risk of default. We believe it has been more effective precisely for the public that still have little or no financial history in the country. Due to the absence of these data, our traditional models tend to be less assertive in the decision for this audience, and in that sense, Worthy Credit has helped by bringing a psychometric characteristic that is independent of history, is able to identify the risk of the operations. We noticed that the public with the best ranking within Worthy Credit presented a default 50% lower in comparison to the other students approved by the standard analysis, which proves the effectiveness of the model.”
How do students respond to these types of assessments?“
At this point, our biggest challenge with Worthy Credit is to engage more and more students to take the questionnaire. In a world where banks and financial institutions have been providing digital solutions with faster and simpler onboarding, requesting the completion of an extra questionnaire can make the student’s journey less attractive, but in the long run I believe that understanding the benefits of providing this information should outweigh the time little extra time it takes to fill it out.”
Can your successful models for student lending be applied to other underbanked populations?
“I believe that the credit modeling that Pravaler does is quite efficient for the audience that specifically seeks student funding. Any new audience or product requires a closer look, since income is something important in the credit decision. Anyway, we have a great opportunity in the future to offer other types of financing to students and their families who have built a credit history with us.”
“The pandemic brought a worsening for all students, including those who were well evaluated. The situation required adjustments to our concession policy to protect the portfolio in this scenario, without harming the access of thousands of Brazilians to higher education… In this way, we were able to maintain our portfolio with very stable delinquency rates throughout the pandemic.”
How has COVID-19 played a role in student lending?
“We observed that the pandemic brought a worsening for all students, including those who were well evaluated. The situation required adjustments to our concession policy to protect the portfolio in this scenario, without harming the access of thousands of Brazilians to higher education. Among the actions we took on credit, we made a major review of predictive models, including the implementation of Worthy Credit, and the cutoff points for approving new students. In terms of collection, we selected students in a more difficult situation to offer payment postponement, in a friendly manner, and we promoted several renegotiation actions throughout the year. In this way, we were able to maintain our portfolio with very stable delinquency rates throughout the pandemic.”
Finally, what does the future hold for credit scoring and student lending?
“Thinking about credit models, our greatest challenge is to expand the use of new data sources (positive registration, digital data, Worthy Credit and Open Banking data) so that negative registration information does not prevail at the time of the decision. In addition, we want to further evolve in the use of “future data” of students for credit decision making today, such as, the probabilities of obtaining employment and consequently the increase in income according to the type of course. In this way, we hope to benefit more and more students with a credit that is healthy for all.”
Amazing stuff! Thank you so much for your time today. The future looks very bright, and we can’t wait to see what is next!