The loan officer who never knocked
You are Felix. You joined a microfinance institution the same week as your colleague Amara, you cover adjacent territories, and your repayment rate is identical to hers. But your territory is rural, clients spread across a 40-kilometre radius, and you spend more time per visit just on fuel and road. Your cost-per-loan is higher. Your portfolio is smaller. At the quarterly review, every dashboard metric the credit committee sees makes Amara look more productive, and nobody says a word to you about it, because nobody has to. The target sheet already said it.
This is the part of financial inclusion that rarely gets discussed. The conversation tends to focus on outreach, on mobile money, on the heroic loan officer cycling into remote villages. What it skips is the machinery sitting above that officer: the incentive structures, portfolio targets, risk-scoring rules, and board-level directives that determine, with quiet precision, which borrowers are worth the trip. Microfinance institution governance doesn't just shape how an organisation is managed. It shapes the geography of credit itself.
How a loan officer's targets become a borrower's ceiling
Start with the most direct mechanism: individual performance metrics. A loan officer at a typical institution might be expected to maintain a portfolio of 200 to 400 active borrowers, keep a portfolio-at-risk ratio below 5 percent, and disburse a minimum volume each month. These are not unreasonable targets. They are, however, filters.
The institution's response, in many cases documented by researchers at CGAP and the Consultative Group to Assist the Poor across several decades of field observation, is to quietly shrink Felix's territory or redirect resources toward denser zones. Nobody issues a memo saying rural borrowers don't matter. Nobody needs to.
This is the first governance mechanism: performance metrics that are indifferent to geography but not neutral in their effect. Indifference, applied consistently, is its own kind of policy.
The credit committee and its hidden model of a good borrower
Above the loan officer sits the credit committee, the body that ratifies or rejects loan applications above a certain threshold. This is where governance gets subtler and, in some respects, more consequential.
Credit committees work from scoring models built on historical repayment data from existing borrowers, which means they encode, structurally, the profile of whoever the institution has already reached. If the institution spent its first decade lending to urban market traders, its scoring model will reward applicants who look like urban market traders: steady daily revenue, fixed business premises, a guarantor with a mobile number and a bank account. Consider the woman running a small dyeing business out of a concrete shed on the outskirts of a mid-sized town. She has customers, a track record, and equipment worth roughly three times what she wants to borrow. She doesn't score badly because she's a bad risk. She scores badly because she's a different kind of risk than the model was trained to assess. Her revenue is seasonal, her premises are informal, her guarantor is a cousin in the same informal economy. The model doesn't know how to value any of that, so the credit committee, acting entirely rationally within its own framework, declines her, or asks for conditions she cannot meet.
The board sets the acceptable portfolio-at-risk ceiling. The credit committee enforces it. The scoring model operationalises it. None of these actors is individually biased. The system is, and the system is the governance.
This is the second mechanism, the one that persists longest: risk models that replicate the past rather than interrogate it.
When donors and investors write the rulebook
A less obvious governance layer sits even higher: the funders. Most microfinance institutions of any scale are capitalised by a mix of development finance institutions, impact investors, and commercial lenders, each bringing covenants, reporting requirements, and sometimes explicit portfolio composition rules.
A development bank might require that 40 percent of a portfolio go to women borrowers. A commercial investor might require that the average loan size stay above a floor protecting their return on administration costs. An impact fund might restrict lending to specific sectors. These constraints can pull in opposite directions, and management has to satisfy all of them simultaneously. If a commercial investor's floor on average loan size is set at, say, 500 dollars in a market where the most underserved borrowers need 80 dollars, the math resolves itself by lending upmarket. The institution drifts toward borrowers who need larger, easier-to-administer loans. The investors get their covenant satisfied. The 80-dollar borrower gets nothing.
This phenomenon, called mission drift in the development finance literature, is not a failure of intention. It is a predictable output of governance structures where capital providers hold different objectives than the institution's stated mission, and where those capital providers have more influence than the borrowers do. Calling it drift is almost too gentle: it is a structural tilt, built in at the founding documents.
The deepest cut: what never gets measured
The most consequential governance failure is the one that produces no data at all.
Loan officers don't file reports on the people they didn't visit. Credit committees don't record the applications that never arrived. There is no metric, in the standard management information systems used by most microfinance institutions, for the depth of exclusion in a given territory. What gets measured is portfolio quality, disbursement volume, repayment rates, operational efficiency. What doesn't get measured is the gap between the institution's stated target population and the people it actually reaches.
Governance responds to measurement, as any student of institutional history could tell you. Boards set targets for what they can see. Investors hold management accountable for what appears in quarterly reports. If exclusion leaves no trace in the data, it generates no pressure for correction. The silence is self-sealing.
Some institutions have tried to close this gap. The Progress out of Poverty Index, developed by the Grameen Foundation, gives loan officers a ten-question survey to assess a borrower's likely poverty level at first contact. Social performance management frameworks, promoted by the Social Performance Task Force, push institutions to track the poverty profile of new clients over time. These tools exist. They are not universally adopted, and in institutions where the board is primarily accountable to commercial investors, they are often the first things cut when operating budgets tighten. Useful, when present. Optional, by design.
The result is a governance architecture highly efficient at optimising for the clients it already has, and structurally incapable of noticing the ones it doesn't.
What people get wrong about fixing this
The instinct, when confronted with exclusion, is to call for more outreach: more loan officers, more mobile channels, more village banking groups. This is the wrong instinct. Outreach without governance reform sends loan officers into new territories operating under the same targets, the same scoring models, and the same portfolio constraints that created the exclusion in the first place. You get the appearance of expansion with the reality of the same filter applied to a wider net. The net gets bigger; the mesh stays the same.
The second common mistake is treating this as a technology problem. Digital credit scoring, proponents argue, can assess thin-file borrowers by using mobile phone data, utility payments, even social network patterns. Some of these approaches have real promise. But the scoring model is still a governance artifact, shaped by who built it, what data it was trained on, what false-positive rate the institution is willing to tolerate, and who on the board decided those parameters. A biased governance structure will produce a biased algorithm. Ask yourself: has any technology in the history of financial services ever corrected a power imbalance it wasn't specifically designed to correct? The algorithm just makes the bias faster and harder to appeal.
Real reform looks less exciting. It means boards that include representatives accountable to borrowers, not just to investors. Performance metrics for loan officers that include a measure of client poverty depth, not just portfolio volume. Credit committees that review not just the applications they received, but the territories they're not hearing from. Funding structures where at least some capital carries patient enough horizons to accept the higher per-loan costs of deep outreach.
None of that is impossible. The governance of Grameen Bank in its original design explicitly tied branch performance to poverty targeting. BancoSol in Bolivia went the other direction as it commercialised, and researchers tracked the resulting upward drift in client income profiles across two decades of data. Both trajectories were driven by governance choices, not by the market or the mission alone. The comparison is almost too neat, but the data holds.
The woman with the dyeing business is not waiting for a technology breakthrough. She is waiting for someone to change the spreadsheet that decides whether Felix's territory is worth the fuel. That spreadsheet is a governance document. Treating it as anything less is the most expensive mistake the sector keeps making.