Latest / Fintech Conversations with Fexingo: Banking, Payments, and Financial Software Companies / How Fintechs Use Real-Time Open Banking Data for Dynamic Credit Card Limits
Transcript
- Lucas: If these conversations are useful for what you're building or running, you're in the right place. Today we're looking at how fintechs are using open banking data to adjust credit card limits in real time. Luna: I've seen this pop up a lot lately — the idea that your credit limit isn't static anymore, but it can change based on your actual financial behavior. Lucas: Exactly. The traditional model is: you apply for a card, the issuer pulls your credit report, gives you a fixed limit, and maybe after six months you get an auto-increase. But a new wave of fintechs is saying, why wait? Why not use live bank data to set a limit that reflects what you can actually afford, and adjust it as your income or spending changes? Luna: So it's less about your FICO score at a single point in time and more about your ongoing cash flow. Lucas: Right. And the poster child for this approach is Petal. They launched a Visa card that uses open banking — with your permission — to connect to your primary bank account. Then they look at your income, your recurring expenses, your spending patterns, and they set a credit limit based on that real data, not just a credit score. Luna: And they can adjust it over time too? If you get a raise or pay down debt, the limit might go up? Lucas: That's the idea. Petal says they can update limits as often as every pay cycle. So if you start a new job or your spending habits improve, the card responds. It's a dynamic credit line that grows with you. Luna: That's a big deal for people who are credit invisible or have thin files. I read that about 26 million Americans are 'credit invisible' — they have no score at all. Open banking data could give them a way in. Lucas: Exactly. Instead of a catch-22 — you need credit to build credit — you can prove your creditworthiness through your actual banking history. Petal reports that about 70 percent of their cardholders were either new to credit or had subprime scores when they started. And their loss rates are competitive with traditional issuers. Luna: That's impressive. How does the risk modeling work differently? They must be using some kind of machine learning on transaction data. Lucas: They are. Traditional credit scores are based on a small set of variables: payment history, utilization, length of history, new credit, mix. Petal looks at hundreds of data points from your bank feed — things like the consistency of your deposits, your spending categories, how much buffer you keep in your account. They built a proprietary model called CashScore that predicts default risk using that real-time data. Luna: So it's almost like they're underwriting your future behavior based on your recent financial habits, not just your past debt management. Lucas: That's a great way to put it. And because they have ongoing access to the data, they can monitor for changes. If someone's deposits start declining or they bounce payments, the model can flag risk early and possibly reduce the limit preemptively. Luna: That sounds smart from a risk perspective, but it also raises privacy questions. Giving a fintech continuous read-only access to your bank account — that's a lot of trust. Lucas: It absolutely is. And that's why the regulatory framework matters. In the US, open banking is still largely happening through screen scraping or through APIs enabled by the Dodd-Frank Act's Section 1033, which gives consumers the right to access their financial data. But there's no uniform standard yet. In the UK, open banking is mandated and standardized by the CMA — that's the Competition and Markets Authority — so fintechs there have a clearer playbook. Luna: So in the US, it's more of a patchwork. But Petal uses Plaid, which is one of the major data aggregators. Plaid connects to thousands of banks and provides a standardized API. Lucas: Right. And Plaid has been working on improving security and transparency. But the model still depends on consumers being willing to share their credentials and trusting that the data won't be misused. Petal says they only use the data for underwriting and don't sell it, but that's a promise, not a regulation. Luna: So the trade-off is access versus privacy. For someone who can't get a traditional card, giving up a bit of privacy might be worth it. But there's also a risk that dynamic limits could feel unpredictable — imagine your limit drops because you had a big medical expense or a slow month freelancing. Lucas: That's a real concern. If the model sees a temporary dip as increased risk, it could reduce your limit at the worst possible time. Petal says they give notice before changes and let customers appeal, but the algorithm isn't perfect. There's also the question of bias — if the model is trained on historical data, it might replicate existing disparities. Luna: That's something regulators are watching. The CFPB has been looking into algorithmic decision-making in credit. And the Federal Reserve's recent report on open banking highlighted both the promise and the need for consumer protections. Lucas: So the space is evolving. Another interesting player is Tally, which uses open banking to manage credit card debt and optimize payments. But they don't issue their own card — they're more of a debt management app. Petal is the purest example of a card issuer using real-time underwriting. Luna: And there's also a company called Kikoff, which offers a credit-building product that reports rent payments to credit bureaus. That's adjacent — using alternative data — but not real-time limits. Lucas: Good point. The dynamic limit model is still niche. Most major issuers like Chase or Capital One haven't adopted it yet. But I think we'll see more experimentation, especially as open banking standards mature and consumers become more comfortable sharing data. Luna: Speaking of comfort — you know, it's interesting. A lot of listeners of this show are themselves in fintech or building products. And this kind of deep dive is exactly why a bunch of them support us directly. Lucas: Yeah, we get that a lot. And it's genuinely helpful. A small group of listeners chips in monthly through buy me a coffee dot com slash fexingo, and that's what keeps these episodes ad-free and focused on real detail. Luna: If you find value in the show and want to help us keep going, that link is simple: buy me a coffee dot com slash fexingo. No pressure, just appreciate you being here. Lucas: Alright, back to the data. Another angle I want to explore is how dynamic limits could change consumer behavior. If your limit can go up when you save more, does that incentivize better financial habits? Luna: That's the hope. Petal's app shows users how their spending and saving affect their CashScore, so it's almost like a nudge. But there's also a risk of gaming the system — people might temporarily inflate their account balances just before a limit review. Lucas: Right, that's why the models look at patterns over time, not just a snapshot. Petal says they analyze at least 90 days of data. So a quick deposit before a limit check won't fool the model. Luna: What about the issuer's side? How does dynamic underwriting affect their capital requirements and risk provisioning? Lucas: That's a great question. Traditionally, credit card portfolios are managed with fixed limits and static risk tiers. With dynamic limits, the risk profile of the portfolio can shift more quickly. That means issuers need more sophisticated risk systems and potentially more capital if limits can increase rapidly. But Petal argues that because they have better data, their overall loss rates are lower, so the net effect is positive. Luna: And they're not the only ones. There's a startup called Self that uses secured cards and then graduates you to unsecured, but that's more of a step function. Dynamic limits are more granular. Lucas: Another emerging model is from fintechs like TomoCredit, which uses cash flow data for underwriting but issues a card with a fixed limit initially. So the real-time adjustment is still rare. I think we're at the early adopter stage. Luna: Do you see traditional banks adopting this? Or will they wait for regulation to force their hand? Lucas: I think we'll see a hybrid approach. Some big banks are already piloting open banking features. For example, JPMorgan has experimented with using deposit data to offer pre-approved credit lines. But fully dynamic limits are a bigger leap because they require real-time decisioning infrastructure. It's more likely that we'll see neobanks and fintechs lead, and then the incumbents acquire or build similar capabilities. Luna: One more thing — what happens in a recession? If a lot of people's incomes drop simultaneously, dynamic limits could tighten across the board, which might amplify the downturn. Lucas: That's a systemic risk that regulators will need to consider. If every card issuer's model simultaneously reduces limits based on similar data signals, it could reduce consumer spending exactly when the economy needs stimulus. It's the same concern with algorithmic lending in general — procyclicality. So far, Petal and others haven't been through a full credit cycle, so we don't know how they'll behave. Luna: So the big picture is that real-time open banking data is enabling a shift from static to dynamic credit. But it's still early, and there are risks around privacy, fairness, and systemic stability. Lucas: Exactly. The technology is here, but the framework — regulatory, behavioral, economic — is still catching up. It's one of those areas where the next five years will be fascinating to watch. Luna: Well, I think that's a good place to leave it. For those of you building in this space, would love to hear your thoughts. Thanks for listening.