API Underwriting and the Future of Lending

Over the past few years, fintech companies like Affirm, Dave, and Fundbox have helped create a new standard of lending.

These companies are different than conventional lenders because they evaluate a borrower’s creditworthiness through their digital footprints. Affirm analyzes social media profiles. Fundbox analyzes Quickbooks accounts. Dave analyzes credit card statements.

This is a unique alternative to the traditional method of assessing a borrower’s risk based on their credit score. Often, credit scores do not accurately reflect a borrower’s true financial profile. This is especially true for younger borrowers who have yet to take out a mortgage, or even a consumer credit line.

As a result, fintech companies are using new data sources to create better predictors of creditworthiness than credit scores alone. Trust is being disaggregated from the credit score companies and being outsourced to APIs. Loans are being underwritten by digital profiles. I predict that this standard in loan-making will expand in a few interesting ways, which I outline below.

On-the-Spot Credit Agreements

Imagine making a purchase at a bookstore.

A retailer selling a particular book for $12 will transfer the ownership of the book to anyone with $12. While the post-purchase settlement might entail splitting the $12 among the retailer, publisher, and author, the transaction has settled for the customer as soon as his (or his bank’s) $12 is delivered to the merchant’s bank account.

In the future point of sale, a bank will make a $12 loan to the retailer on the basis of the buyer’s collateral. When the buyer’s monthly statement is due, he will have to pay back his principal and interest, but this is where it gets interesting.

Let’s image that the book buyer has 20,000 followers on Instagram (I don’t know any software engineers with this many Instagram followers), and three outstanding freelance offers on his Upwork account (he must be a front end guy). Before he goes to purchase his book, his bank will have arranged some collateral on his behalf. This collateral might include API keys to his Upwork and Instagram accounts, as well as the write access to his PEO account.

The buyer will have the option to make his credit payment traditionally, using cash (fiat or crypto), or by signing a smart contract that binds his private-key verified identity and his authenticated API keys to a specific task he is obligated to perform in the future.

Since he has a lot of Instagram followers, he might have the option to make a sponsored post on behalf of a clothing brand, which will earn him his $12 so long as the post generates at least 1,000 impressions.

When his credit statement is due, the smart contract will evaluate the completion of this task by querying his Facebook ad account (curl /<ad_accont_id/insights) and pass the result into the contract ABI.

If the result is True (he completed the task), his credit statement will settle, any interest will be tallied, and the cycle will continue. But, if the result is False (he deleted his Instagram account and never posted what he was supposed to), the contract will call his wallet contract and ask for payout in the statement amount. Since he has already given the bank’s smart contract permission to call for payout on his wallet contract, the bank can automatically deduct the minimum payment, and even charge a late-fee.

I believe this is very likely to happen once banks (1) adopt auto-reconciling smart contracts to issue loans, and (2) develop oracle services which monitor specific REST APIs to determine whether a debtor has settled a loan. Today, money is used as a proxy for intellectual capital. The software engineer’s knowledge is priced into his salary. And while using a proxy for intellectual capital (labor or knowledge converted to cash) is a simpler method of exchange, smart contracts grant us access to a new platform which can issue loans on the obligation of performing complex specialized tasks in the future.

Predictions for 2022

  • Fintech companies will continue to expand the number of touchpoints used to analyze consumer credit worthiness by accessing digital profiles, such as social media and bank accounts, through an API which the user grants access to.
  • Banks will develop digital collateralization platforms to securitize previously unsecured debt. Consumers will have the option to provide a bank with their API keys to the services they control, and these API keys will serve as loan collateral.
  • Banks will eventually create cooperative loan programs. Just as government employees are relatively safe borrowers, so too will be the software engineer whose APR is tied to how many commits he makes to a Github repo.

[1] As a parenthetical, it seems likely that PEOs (such as JustWorks) will serve employment-backed consumer loans in the future. A PEO (Professional Employer Organization) platform is an intermediate interface between an employee’s bank and their employer’s bank, and is thus suited to serve loans that are insured by a debtor’s employment.

This introduces an important premise: if capital is sufficiently abundant to reduce consumer reliance on centralized lenders, then loan insurance (not access to capital or market-making) will become a lender’s main value-add.

Did you enjoy this?

If you've enjoyed reading anything here, you should consider signing up for semi-regular emails from me. That's the first place I go to share anything that's on my mind.