n.V1-7.01 | THE SOLUTIONS, AND WHAT THEY MUST PROVE

01| There is no shortage of answers to poverty. Cash, credit, training, clinics, block grants, vouchers, subsidised firms, community meetings — each arrived with a theory, a constituency and a decade of expansion. What is short is the ability to say which of them does anything, and the honest reason is not that nobody has looked. It is that the field has never agreed on what a solution is being asked to do.

02| The usual way of asking is a league table. Models are listed down one axis, criteria across the other, scores in the cells, a winner at the bottom. The table is the weakest part of the exercise. Its result is fixed by two decisions taken before any evidence is read: which criteria enter the grid, and how they are weighted. Add "compounding effect" and every model producing a one-off gain drops. Add "proven at scale" and every untested model drops. Neither addition is a finding. Each is a definition wearing a number.

03| So the criteria are the object of study, not the instrument. In the chapters that follow, each solution is put through four questions. They are not our invention. They are the four places the evidence keeps breaking.

04| Does it compound, or does it decay? Compounding means the gap between treated and untreated widens with time. It is asserted constantly in prospectus language and tested rarely, because testing it requires a panel running a decade past the intervention. Where those panels exist, the usual finding is convergence. Uganda's Youth Opportunities Program handed out roughly $400 a head in 2008. At four years, grantees worked 17 percent more and earned 38 percent more. At nine:

"we find these gains have dissipated. Grantees' investment leveled off; controls eventually increased their incomes through business and casual labor; and so both groups converged in employment, earnings, and consumption levels."— Blattman, Fiala & Martinez (The Long-Term Impacts of Grants on Poverty: Nine-Year Evidence from Uganda's Youth Opportunities Program, AER: Insights, 2020, Vol. 2, No. 3, pp. 287–304)

05| Read that carefully, because the mechanism is not the one people assume. The treated group did not collapse. The untreated group caught up. The programme bought its recipients about five years of earlier arrival at a place they were mostly going to reach anyway. That is a real benefit and it is not the benefit that was claimed.

06| Does it survive a change of implementer? Almost every result in this field was produced by an organisation that chose the intervention, believed in it, hired for it and supervised it. Almost every deployment is run by an organisation that did none of those things. The cleanest test holds the intervention fixed and varies only the carrier:

"New teachers offered a fixed-term contract by an international NGO significantly raised student test scores, while teachers offered identical contracts by the Kenyan government produced zero impact."— Bold, Kimenyi, Mwabu, Ng'ang'a & Sandefur (Experimental Evidence on Scaling Up Education Reforms in Kenya, Journal of Public Economics, 2018, Vol. 168, pp. 1–20)

07| Identical contracts. Teacher characteristics explained almost none of the gap. Whatever was doing the work was not in the intervention at all. This finding should be read before every other finding in this book part, because it means an effect size is a property of a pairing — this thing, done by these people — and not a property of the thing.

08| Does it beat cash at equal cost? The field has quietly solved the comparison problem in a way nobody planned. Rather than agree on a common unit of good, it uses the simplest intervention as the yardstick and asks whether anything else beats it for the same money. That test has now been run against real programmes, and cash keeps winning more often than the sector's spending pattern would suggest. Any solution that has not faced it has not been priced.

09| Who ends up holding the decision? This is the question the sector talks about most and measures least. Participation, ownership, empowerment — all three are measurable, and when measured they return null more often than not. That is not an argument against them. It is an argument against counting them as achieved because a meeting happened and a minute was signed.

10| One more thing to fix before the catalogue begins: the money. It is common to hear that a couple of hundred billion dollars a year is spent fighting poverty and spent badly. That figure is roughly total official development assistance across all sectors — $212.1 billion from DAC donors in 2024, 0.33 percent of their combined national income, down 7.1 percent in real terms and the first fall in six years. Anti-poverty money is overwhelmingly not that. It is domestic social protection budgets, an order of magnitude larger, allocated through national politics and touched by no donor. A critique aimed at the aid budget is aimed at the small, shrinking minority of the money.

11| Two decades of trials have produced a literature that is far better at saying what does not work than the sector's brochures admit, and far more honest about its own limits than its critics allow. The microcredit verdict is the model of how this goes:

"a consistent pattern of modestly positive, but not transformative, effects"— Banerjee, Karlan & Zinman (Six Randomized Evaluations of Microcredit: Introduction and Further Steps, AEJ: Applied Economics, 2015, Vol. 7, No. 1, pp. 1–21)

12| Modestly positive but not transformative is, on the evidence, the honest summary of nearly everything in this book part. The chapters that follow are an attempt to say why, one solution at a time, and to name the few places where something more than modest has actually been demonstrated.