n.V2-3.03| THE COMPOUNDING LOOP

01| Most anti-poverty work follows the same arc. The programme launches and delivers. Results appear. Funding continues for a few years and impact grows roughly in line with the money. Then the grant ends, and within two or three years the measured gains have drifted back toward where they started. Cash transfers raise consumption while the payments run. Job training raises employment until the skills go stale. New wells and roads improve conditions without touching the dynamics that produced the conditions.

02| The sector has largely accepted this. It shows up in the language: programmes are designed with "sustainability plans," which on inspection usually mean plans for how to keep the funding going forever. The question almost nobody asks is different and harder. What would make the impact continue after the money stops?

03| A small number of interventions behave differently. They start slower, look less impressive on year-two metrics, and then keep growing as external funding falls away. The difference between the two patterns is not management quality, budget size, or technology. It is structural, and it comes down to which of two mechanisms the intervention is built on.

04| The first mechanism is resource provision. Its logic: people are poor because they lack something — money, food, skills, a clinic. Provide the missing thing and poverty falls. Stop providing it and poverty returns. Impact is a function of investment; when investment goes to zero, impact goes to zero. This is not a criticism of the people running such programmes. Within its own logic it works exactly as designed. It provides the fish. What it does not do is change the person's capability to catch fish, their belief that they could, their knowledge of where the fish are, or their access to others who fish.

05| The second mechanism is capability building with network effects. Its logic: people stay poor because they lack the framework of capabilities to generate and sustain progress themselves. Build the capabilities and people generate their own progress; early success builds further capability and visibly inspires others; the effect spreads. Impact becomes a function of investment and time and network. When investment falls, impact can keep growing, provided the network effects are strong enough to carry it.

06| Four things make the second mechanism compound where the first decays. Capabilities build on capabilities — a small success produces the confidence to attempt something larger, which produces a larger success. Success is visible and it transmits: when a neighbour who is demonstrably not exceptional succeeds through her own effort, the belief about what is possible in that street changes. Capable people create value for others, so each new capable person raises the ceiling for everyone around them. And as the number of capable people rises, the value of being connected to them rises with it.

07| Resource provision creates dependency by construction. Capability building creates agency by construction. That is the whole argument, and everything IkoCiti does downstream is an attempt to be reliably on the right side of it.

08| The honest complication is timing, and it is the part funders most need to hear. Compounding does not begin on day one. There is an activation threshold, and the arithmetic before it is unimpressive.

09| In the first three years, growth is essentially linear. Every person developed requires direct intervention; ten people reaching intermediate capability might inspire two or three others to start. Between years three and six, the first cohort reaches the level where they develop others, and growth rate roughly doubles per cycle — weak compounding, visible if you are looking for it. Between years six and ten, a critical mass of experienced people are each developing three to five others while network effects add their own multiplier, and the trajectory turns clearly exponential. Past year ten, the system reproduces itself and growth is limited mainly by the size of the population.

10| The implication is uncomfortable and specific: you have to fund through the threshold. A well-designed intervention funded for two or three years and then released will show linear impact that decays, not because it was badly designed but because the money stopped before the multiplication started. The same intervention funded for seven years can become self-perpetuating — not because the cheque was larger, but because it lasted past the point where the system began generating its own growth.

11| This reframes what a funder is actually buying. Short grants buy outputs. Long grants, into the right structure, buy a mechanism. The sector's standard two-to-three-year cycle is almost perfectly calibrated to stop just before the interesting part.

12| We are not claiming to have discovered compounding. Peer effects, social proof and capability progression are all well documented. What we claim is that almost nothing in the anti-poverty sector is designed for them — that viral spread is treated as a happy accident rather than a design requirement, and that measuring outcomes without measuring the mechanisms that produce compounding leaves the sector unable to tell the two patterns apart until it is far too late to matter.