Alternative data could unlock bank loans for Philippine MSMEs

Market vendors, small retailers, and logistics operators across the Philippines may have steady cash flow, but many still cannot get a bank loan because they lack the formal documents lenders demand, a fintech executive said.

Allen Xavier, Philippines country manager of AND Solutions, said alternative-data underwriting can turn bank statements, invoices, handwritten records, and transaction histories into usable credit signals for micro, small, and medium enterprises (MSMEs).

Bank lending to MSMEs reached PHP 574.8 billion as of end-December 2025, according to the Bangko Sentral ng Pilipinas (BSP), accounting for only 4.73 percent of the banking industry’s total loan portfolio.

The figure was up 5.23 percent from PHP 546.22 billion a year earlier, against a total industry loan book of PHP 12.143 trillion.

Micro and small enterprises, the segment Xavier described as most misjudged, received PHP 238.45 billion, or 1.96 percent of all bank loans, while medium enterprises got PHP 336.35 billion, or 2.77 percent.

The gap matters because MSMEs make up nearly all registered businesses in the country, and a lender that cannot read their creditworthiness leaves them with few options beyond informal moneylenders whose steep rates eat into the working capital that keeps stalls, stores, and delivery services running.

Better credit assessment, Xavier argued, could move these businesses into regulated finance, where borrowing costs are lower, repayment builds a credit history, and access to insurance and savings products follows.

The Magna Carta for MSMEs once required banks to set aside 10 percent of their loan portfolios for the sector, 8 percent for micro and small enterprises and 2 percent for medium enterprises, but that mandate expired on June 16, 2018.

Documentation or inertia?

Asked whether the lending gap is a documentation problem or reluctance on the part of banks, Xavier said: “It is a combination of both, but structural inertia in risk modeling carries a heavier weight.”

“While the lack of formal documentation like audited financial statements or BIR tax returns creates an immediate barrier, traditional banks remain tethered to collateral-based, asset-heavy risk models,” he said.

“Furthermore, hesitation from the Banks come primarily from updating these frameworks which requires significant investments in technology, as well as data integration and regulatory approvals,” Xavier said.

“For many traditional institutions, the high cost of underwriting small-ticket micro-SME loans under legacy procedures simply makes the segment appear high-risk and unprofitable, leading to a fundamental reluctance to adapt,” he added.

Xavier said the underserved segment is also a business opportunity.

“On the other hand, 5% of loan portfolio for MSMEs which comprise 99.63% of all registered business establishments in the Philippines, according to data from the Department of Trade and Industry (DTI) and the Philippine Statistics Authority (PSA) is still a huge potential market segment that can be explored for someone willing to focus on that segment with the right risk appetite,” he said.

What banks look for

“Standard models heavily over-rely on formal financial metrics that rarely exist in the informal economy,” Xavier said.

On audited financial statements and tax returns, he said: “Informal businesses often keep cash-based, informal ledgers rather than standardised accounting reports.”

On credit bureau scores, Xavier said: “Many micro-SME owners have limited or no formal credit history, yielding empty or misleading credit files.”

He said reliance on real estate, heavy equipment titles, and other acceptable assets as hard collateral “excludes businesses whose primary assets are daily cash flow, inventory turnover, or vendor trust.”

On operating history and bank balances, Xavier said: “Micro-SMEs frequently reinvest cash daily or use personal accounts for business transactions, making stationary bank balances a poor proxy for actual revenue.”

Reading handwritten ledgers

AND Solutions uses AI-powered optical character recognition (OCR), technology that converts images of printed or handwritten text into machine-readable data, to process handwritten records and invoices.

Asked how the system copes with the wide variation in how vendors and small retailers keep their books, Xavier said: “The system handles variance through a multi-layered computer vision and natural language processing pipeline.”

On pattern recognition, he said: “Instead of relying on rigid templates, the AI uses localised machine learning models trained specifically on specific documents typically used by lending as a basis for supplemental documentation.”

The system also checks entries for consistency. “The engine cross-checks recognised entries against operational logic, matching stated unit prices with current local market benchmarks, inventory turnover rates, and batch tallies,” Xavier said.

“With ample training effort, the AI models can be taught to learn semantic data structures from Raw text from certain documents like: handwritten logbooks, tindahan notebooks, or delivery receipts,” he said.

“These data can then be mapped into standardised financial categories (e.g., daily sales, supplier payouts, inventory costs),” Xavier added.

On fraud, he said: “Algorithmic sanity checks flag inconsistent entries, improbable volume spikes, or altered figures for human verification before outputting a confidence score.”

“Our solution features such as cross-checking allows analysts to be flagged when documents submitted have inconsistent information across submitted documents; metadata analysis allows the system to monitor documentation authorising and tampering wherein generally the data footprint of submitted documents can be captured,” Xavier said.

Better predictors of repayment

Beyond digitized notes, Xavier said: “The strongest predictors leverage real-time operational flow rather than static assets.”

He cited digital wallet and QR Ph transactions, referring to the national QR code payment standard. “Frequency, consistency, and volume of daily digital merchant payments (e.g., GCash, Maya) directly reflect active cash flow,” he said.

On supplier and logistics settlement data, Xavier said: “Consistency in paying distributors or clearing delivery orders demonstrates operational stability and reliable commercial relationships.”

“Timely load top-ups, mobile data usage patterns, and utility bill settlements act as reliable proxies for baseline financial discipline,” he said of utility and telco payments.

On e-commerce and point-of-sale records, he said: “Order fulfillment rates, return ratios, and customer review scores on digital marketplaces showcase business viability and operational efficiency.”

Moving away from ‘5-6’

Many unbanked MSMEs borrow from informal lenders under the “5-6” system, so called because borrowers typically repay six pesos for every five borrowed over a short period.

“The transition relies on a combination of multiple factors, but one good approach is to start with a foundational principal approach like frictionless onboarding and progressive risk profiling,” Xavier said.

On onboarding, he said: “Alternative underwriting replaces lengthy document submissions with automated background checks using existing digital traces (e.g., e-wallet histories or supplier receipts).”

“Merchants enter the ecosystem with small, manageable loans tailored to daily cash cycles, mirroring the familiar repayment structure of non-traditional, often cost-prohibitive sources of credit without the predatory interest rates,” he said.

“Institutions in order to effectively monitor this should incorporate more information related to the micro loans such as purpose and loan durations,” Xavier added.

On tiered electronic know-your-customer (e-KYC) checks, he said: “Financial institutions can apply risk-proportionate identity verification, enabling basic access with simple valid IDs and expanding limits as the relationship matures.”

“Integrating micro-repayments directly into daily supply chain purchases or digital wallet checkouts makes compliance seamless and natural,” he said of automated embedded repayments.

On revolving credit facilities (RCF), Xavier said: “Offering new loan types like: RCF allows businesses use to manage daily cash flow and short-term expenses while establishing continuous relationship and build up of credit history.”

Building resilience

Once a micro-SME joins the regulated system, Xavier outlined a four-stage path toward long-term financial resilience.

The first is establishing a credit footprint. “Successfully repaying initial micro-loans creates a formal credit history within regulated credit bureaus and partner financial institutions,” he said.

The second is larger working capital. “As transaction history builds, credit limits automatically scale, allowing the business to fund larger bulk inventory purchases, smooth out seasonal dips, or secure better supplier pricing,” Xavier said.

The third is product diversification. “Access expands beyond short-term working capital into business insurance (e.g., crop or inventory protection), merchant lines of credit, and yield-bearing savings accounts,” he said.

The fourth is formalization. “Ongoing interaction with digital financial tools leads to natural digitisation of records, better cash-flow management, and eventual readiness for standard bank products,” Xavier said.

Reaching the provinces

On adoption outside Metro Manila, where vendors may lack digital literacy or reliable connectivity, Xavier said: “There are many approaches to address adoption bottlenecks and there is no ‘t-shirt sizing’ or best practices that can meet everyone’s scenarios.”

He pointed to agent-assisted, or “phygital,” onboarding. “Partnering with local field officers, cooperatives, or supplier distributors allows agents to capture images of physical ledgers or assist with e-KYC on behalf of the vendor,” he said.

“While field presence remains an important component of our initial touchpoints, this serves as a foundation for our broader digital transformation strategy, designed to build trust and readiness for fully automated services,” Xavier added.

On offline-first processing, he said: “Mobile applications designed for field deployment store data locally during connectivity outages and automatically sync with central AI underwriting engines once a connection is restored.”

“Tools optimised for basic smartphones, simple messaging platforms (e.g., SMS, Viber), or voice-guided interfaces lower technical barriers for users with low digital literacy,” he said.