What If We Got Lending Right in Mexico?

Credit is one of the greatest engines of economic value creation. It allows individuals to improve their lives, families to access stability, and businesses to grow beyond what their cash flow can support. But in Mexico, lending isn’t a calculated risk, it’s an ultra-high-stakes gamble. And far too often, the house loses.

In a market where trust is fragile, institutions are weak, and income is uncertain, even the best intentions can collapse into default. I’ve spent years in the trenches of lending, from developing installment-based Buy Now, Pay Later (BNPL) platforms, to merchant cash advances through Point-of-Sale (POS) systems, to structuring loans for oil & gas suppliers backed by Pemex contracts. What I’ve seen is this: technology can help, but it cannot substitute for a deep understanding of the people you’re lending to.

The Reality on the Ground

Most traditional banks in Mexico focus their credit portfolios on the safest segments: large corporates on one side, and high-FICO-score credit card customers on the other. These are the easiest to model and manage, predictable, compliant, and profitable.

Fintechs and neobanks have stepped into the space left behind, trying to serve underserved or subprime segments with new tools, but most are losing money. That’s concerning because we need them to succeed if we want a truly inclusive financial system. And to top it all off, the largest SOFOMs, Unifin and Credito Real, went bankrupt in 2022. But in Mexico, the risk of default is built into the foundation, and if we don’t name the root causes, we can’t fix them.

Why People (and Businesses) Don’t Pay

In my experience, defaults stem from three core issues. Insufficient income: most people and small businesses operate with little financial cushion, and a single unexpected event, illness, job loss, macroeconomic shock, destroys their ability to pay, which isn’t irresponsibility, it’s survival. Weak credit culture: creditworthiness is not a deeply embedded value, many borrowers don’t monitor their credit score, understand its impact, or see non-payment as a moral failing, and the system doesn’t incentivize credit discipline strongly enough. Limited legal enforcement: if a borrower chooses not to pay, legal recourse is nearly useless even when assets are involved, the judicial system is too slow and inefficient, and with the recent judicial reforms creditors may be even more exposed in the future.

That means we’re trying to build lending models in a context where both ability and willingness to pay are fragile. Where data is incomplete, and enforcement is weak. And no AI in the world can change that.

Tech Can’t Solve What Culture and Policy Ignore

Over the last decade, there have been big advances in areas like KYC and onboarding, eliminating what used to be long and manual processes, and also minimizing fraud like identity theft.

But when it comes to measuring the actual ability to repay a loan in the Mexican context, the correct tools are still missing. Startups have tried to plug these gaps using behavioral signals: analyzing social media profiles, education levels, even psychometric tests. Some of these efforts are intellectually interesting. But when cash flow is unstable, and incentives are misaligned, they simply don’t work at scale.

In my experience, two things still matter more than any advanced model: FICO scores and real expected cash flow.

That’s why our merchant cash advance platform worked, not because it had a flashy algorithm, but because we controlled the cash flow. The loan was based on real-time sales data from POS terminals. We debited repayments before the merchant ever touched the money.

Similarly, when we financed SMEs in the oil & gas sector, our strategy was to fund only those suppliers with formal Pemex contracts, and to take assignment of the payment rights. As long as the work was completed, we got paid. In this case the SME didn’t turn out to be the risk, it was Pemex. And when Pemex stopped paying suppliers, even our airtight structure couldn’t protect us. This model worked not because of software, but because of smart underwriting. And it required real verification. Not just APIs, but field visits, phone calls, and eyes on the contract.

Building a Lending System That Works

So how do we improve this? First, we have to acknowledge that lending alone doesn’t fix poverty. When borrowers don’t earn enough, they will default, period. That means we need to pair lending with structural solutions.

Design for volatility: create loan products with embedded flexibility, like income-based repayments, automatic grace periods, or emergency buffers. Invest in financial education: make credit scores matter by showing people how their financial behavior affects their access to future opportunities, and create real incentives for repayment. Build shared data infrastructure: we need universal, interoperable credit histories that go beyond traditional credit bureaus, including open banking, POS sales data, and even invoice payments. Enable trust-based enforcement: the courts won’t save us, but maybe platforms can, so imagine a lending tool where suppliers upload their contract progress, and buyers like Pemex verify completion directly in the system, giving you transparency and leverage. And design for Mexican reality: most importantly, we need tools built for Mexico, not imported from Silicon Valley, technology that fits the way we live, work, and relate to money, simpler, culturally fluent, and trustworthy.

In time, we could turn a broken system of distrust, high risk, and high rates, into a virtuous system of trust, low rates, growth, and prosperity.

Finance Meets Humanity

Ultimately, we need lenders who combine financial rigor with human empathy. Who understand spreadsheets and the rhythm of the street. Who can model portfolio risk, and verify a purchase order with a field visit. Who can design scalable platforms that don’t just collect data, but build trust.

The problem of lending in Mexico won’t be solved with code alone. It requires design, discipline, fieldwork, and imagination. And if we can crack this, the opportunity isn’t just to create better credit, but to unlock the full economic potential of millions of families and businesses.

We don’t need easier money, we need smarter money, grounded in reality, scaled with tech, and guided by people who care. That’s the kind of work I want to do, and that’s the future I want to help build.