Zoom in on the number of upstart variables to analyze the well-being of American consumers – 05/10/2022 at 08:46

Flash News ODDO BHF Artificial Intelligence April 2022

Stuck between the risks of inflation and the economic downturn, the health of the world economy should again depend on American consumers. The latter, as we know it, calibrates its use by measuring the ability to access credit. This is the context that inspired us to show the revolution enabled by artificial intelligence in American consumer credit scoring; The main social benefit of allowing credit to consumers is that those who in the past were excluded from it due to the limitations of the traditional model for assessing their well-being.

The limits of the US consumer credit rating model …

Credit access conditions for American consumers are governed by higher formalities than what we encounter in Europe. Most banks and institutions that lend use a FICO score: a 3-digit score (between 300 and 850) in the hands of the exclusively listed company Fair Isaac Corporation (FICO US). These scores, dating back to 1989, are calculated on the basis of third party data provided by three national credit reporting agencies (TransUnion, Equifax and Experian). These are typically based on data such as your borrower’s history (regularity of your repayments, combination of your debts, etc.) and the total amount of your commitments. They form a file analysis column for consumers who request new credits for risk analysis of already subscribed credits.

These traditional models for assessing consumer creditworthiness have now reached the limit of order:

  1. Social (A good portion of mennials today do not use credit cards and so they have not made a history of borrowing when they can still decide to save as a result of being a borrower with good power repayment).
  2. Overall (Analysis of the well-being of American consumers in the twenty-first century will require the incorporation of far more parameters than the parameters initially provided by the FICO score).
  3. Financial (Traditional FICO models focus on consumers’ ability to repay their debts on time rather than producing the consumer’s full cash flow when, for example, it is possible to repay their debts with other credits).

The ultimate goal is to reinstate more equity in the system so that good borrowers stop paying unwanted premiums on their credit costs and, conversely, more suspicious debtors can continue to access credit.

… Defeated by artificial intelligence today

In a world where 80% of American consumers have never defaulted on their maturity and yet only 48% of them access credit at an optimal price, FinTech introduces a technical hurdle in the face of upstart traditional valuation methods. American consumer indebtedness. It uses the power of artificial intelligence to better assess the indebtedness of American consumers. In the last eight years, Upstart has created a dataset consisting of 1,500 variables (compared to about 20 for FICO scores). Artificial intelligence algorithms have been trained in more than 20 million redemption sequences; As a result of being able to establish rules and relationships that did not exist in the past; This creates a barrier to entry against Upstart’s competitors. Indeed, it would be inappropriate to use thousands of data to make credit decisions without the existence of sophisticated machine learning models.

This allows Upstart to introduce credit analysis criteria (which in most cases do not exist in the FICO score) such as the nature of the job you hold, your academic background, your banking transaction history, your standard of living and even interactions with your consumer credit sites. In short, artificial intelligence allows a very holistic approach to the well-being of American consumers, in the service of greater fairness in terms of everyone’s borrowing power.

Upstart’s Artificial Intelligence platform has been made available to banking and credit institution partners through an application available in the cloud. Customers can access the Upstart Appraisal Service (upstart.com) on their behalf through the company’s website or through the partner bank’s website.

Read the full article in PDF below.

Past performance is not a reliable indicator of future performance and is not constant over time. None of the companies mentioned above recommend investment.

Leave a Comment