One pension scheme reached eighty million people. Another reached a hundred thousand. Both miss the smallholder farmer.

Zambia has just given itself a new legal shelf for informal sector pensions. The National Pension Scheme Act, 2026 creates what the reform's own guidance calls “ring-fenced pension or savings arrangements established by statutory instrument to cater for specific groups such as: Diaspora Zambians, Informal sector workers, Voluntary contributors,” and it lets a person register before they have secured formal employment at all.¹ It is a real opening. The question worth asking before anyone celebrates it is whether a shelf was ever the missing piece.

The gap, in outline

Across sub-Saharan Africa, social protection coverage has moved from 15.2 percent of the population in 2015 to 19.1 percent in 2023, and the International Labour Organization is direct about what that leaves: in low income countries, which is most of the continent, not one has a comprehensive social protection system.² Globally, only 33.8 percent of the working age population is even legally covered by a comprehensive social security system, before the separate question of whether people who are legally covered actually contribute.² An earlier ILO accounting, now over a decade old and worth reading as a floor rather than a current figure, put effective old age pension coverage in sub-Saharan Africa at 5.9 percent of the working age population, with 8.4 percent of the labour force contributing to a scheme at all.³

The reason given every time is the same word: informality. Zambia's own FinScope survey illustrates the problem inside the number. Pension access rose from 6.3 percent of adults in 2020 to 14.4 percent in 2025, which reads as real progress until the survey notes that the two figures were built differently: 2020 asked who contributes to a pension, 2025 counted adults in formal employment.⁴ Some or all of that improvement may be a change in the ruler, not a change in what it measured. That is a contradiction worth sitting with rather than smoothing over, and it is the kind of thing that shows up only when two vintages of the same national survey are read side by side rather than quoted separately.

Two designs, two very different outcomes

The standard explanation for low informal sector coverage is that a pension built for salaried life, a fixed sum on a fixed date, does not fit a life where income arrives unevenly. The policy response, tried in several countries over the past fifteen years, has been to make the contribution flexible: no fixed amount, no fixed date, pay what you have when you have it.

Kenya's Mbao Pension Plan is the clearest test of that idea. Launched in 2009 for the roughly twelve million members of the Jua Kali informal trade associations, it set a minimum contribution of KES 20 a day, about USD 0.19, payable whenever the saver chose through mobile money, with no penalty for missing a day.⁵ By 2018, after nine years, it had reached 100,000 members and a fund of about USD 1.34 million, a coverage rate under one percent of its own target population, and the average member was contributing only about KES 180, roughly nine days worth of the minimum, in a typical period.⁵ The scheme's own reviewers point to a second, sharper problem: because withdrawals face no real restriction after an initial three year window, many members treat the account as emergency savings and empty it at the first shock, which defeats the purpose of a pension without technically breaking any rule of the product.⁵ Flexibility solved the affordability problem it was built for and created a liquidity problem nobody had designed against.

India's Atal Pension Yojana is close to the opposite design. Contributions are fixed, not flexible: a set amount taken monthly, quarterly or half yearly by automatic debit from a savings account the subscriber must already hold, in exchange for a government guaranteed pension of between INR 1,000 and 5,000 a month from age 60.⁶ For its first five years the government matched half the subscriber's contribution up to INR 1,000 a year for non-taxpayers.⁶ At its tenth anniversary in 2025, the scheme had passed eight crore, about 80 million, enrolments, with 3.9 million more added in that financial year alone.⁶

That is the anomaly. The design everyone assumes should fit informal, uneven income, flexible and forgiving, is the one that stalled. The design that looks least suited to it, fixed and automatic, is the one that scaled by orders of magnitude.

The variable that actually explains it

Read past the headline and the two schemes were never really testing the same variable. Atal Pension Yojana's scale did not come from asking less of the saver. It came from attaching the contribution to money that was already moving through a bank account on a schedule the saver did not have to remember, backed by a government guarantee and, for five years, a matching subsidy.⁶ Mbao asked the saver to remember, to choose an amount, and to resist withdrawing, with no rail underneath the decision and no guarantee behind the outcome.⁵

The lesson is not flexible versus fixed. It is whether the contribution rides on an event that already happens reliably, or depends on a fresh act of will every time.

This is where the finding stops being about informal workers in general and starts being about smallholder farmers specifically, because a smallholder is the case where neither existing design's underlying event holds. Auto-debit from a bank account assumes money is landing in that account often enough to be debited from, which fits a market trader or a bus driver earning most weeks of the year far better than it fits a farmer who is paid once or twice a season. A flexible daily top-up assumes the saver has small discretionary sums arriving often enough to make a habit of it, which is closer to true for casual urban labour than for a household whose cash is concentrated at harvest and thin for months either side of it. Neither the Indian model nor the Kenyan one was built around a single, large, seasonal cash event; both assume some version of a drip.

Zambia has direct experience of this gap. In 2018 the National Pension Scheme Authority signed memoranda with seven informal sector associations, domestic workers, drivers, saw millers, marketeers and traders among them, to extend voluntary coverage, and ran a parallel initiative named SPIREWORK with the ILO and Finland's National Institute for Health and Welfare aimed specifically at small scale farmers and rural agriculture workers.⁷ Seven years on, no coverage results for that specific initiative appear in the public sources checked for this piece. Whether it reached a small number of farmers, was folded into a wider programme, or simply was not the subject that got published is not established here, and is left as an open question rather than a verdict.⁷

What the new Act does and does not resolve

The 2026 reform gives NAPSA the legal room to build a ring-fenced scheme for informal sector and voluntary contributors, keeps the contribution rate unchanged at 10 percent shared, and raises the guaranteed minimum pension to ZMW 2,327.¹ What it does not specify, at least in the guidance published so far, is which event a smallholder's contribution would attach to.

If the design defaults to a calendar schedule, whether flexible in the Mbao style or fixed in the Atal Pension Yojana style, the evidence above says it inherits the same mismatch that has limited both. If it is built instead around the moment a farmer's crop is sold, the one point in the year when cash reliably exists and, where a buyer already records the transaction, is already being counted, the design would be doing for a pension contribution what Atal Pension Yojana's auto-debit does for a monthly saver: removing the need for the saver to act from memory or discipline. Whether that specific design choice would work at scale for Zambia's smallholders is not something the evidence gathered here proves. It is DNC's own reading of what the comparison implies, offered as a hypothesis for the people building the scheme to test, not as a demonstrated result.

Where DNC sits in this, and where it does not

DNC's own model already anchors a financial obligation to the harvest sale rather than to a calendar date: a lean season loan is repaid in grain at the point the crop is sold, and every purchase and loan is recorded against the individual farmer.⁸ That is a live example of a contribution style tied to a seasonal event rather than a drip, and it is offered here only as an illustration of the mechanism, not as evidence that the mechanism has been proven at pension scale. It has not.

The honest limit of DNC's own experience is that harvest-anchored recording only works where an intermediary is already standing at the point of sale to capture it, and, on a 2019 World Bank diagnostic, under 5 percent of Zambia's small and medium scale farms had access to any institutional credit at all, under 3 percent to formal sector credit specifically, which is a proxy for how few of those farms sell through a channel that keeps any formal record of the transaction.⁹ Building a pension contribution onto a sale event first requires the sale event to be recorded by somebody, which is a harder and earlier problem than the pension design itself, and DNC's own two trial seasons cover a small fraction of one province, not a basis for claiming the model generalises.