Banerjee and Duflo see poverty as thousands of small market failures

1

Beyond the aid war

The grand fight over whether aid helps or hurts is a distraction because poverty is thousands of separate, testable market failures, not one problem.

Sachs held that poor countries sit in a poverty trap and need a large coordinated push of foreign aid to escape. Easterly answered that aid creates dependency, props up bad governments, and rarely reaches the people it claims to help. Banerjee and Duflo call this a false choice built on historical narratives rather than close evidence from inside households.

Their substitute is the clinical trial: stop asking whether aid in general works and ask whether a specific bed-net program, teaching method, or savings account changes behavior. Compare a treatment group with a similar untreated group. The method was already in use in Kenya in the 1990s; by the early 2000s the authors had built a lab at MIT to scale it.

The core claim running under every later chapter is that the poor are not fundamentally different in their reasoning. They simply operate with far less margin for error, far less information, and far fewer working institutions. Interventions therefore succeed or fail on small design details, not on whether one is philosophically pro-market or pro-state.

In practice: Instead of debating aid writ large, the authors ask whether a particular immunization camp, contract-teacher hire, or lockbox changes measured behavior against a control group.

2

The appetite S-curve

Extra income for poor households does not mainly buy more calories, which undercuts the idea that poverty is fundamentally a nutrition trap.

A popular theory says too little food yields too little energy to work, which yields too little income to buy more food, so a calorie subsidy should break the cycle. Household spending data from India, Indonesia, and elsewhere, building on work by Deaton and Drèze, point the other way. When poor households get extra money, calorie intake rises only slightly.

Spending shifts toward more expensive, better-tasting food—white rice instead of cheaper grains, more sugar, tea, and festival meals. Households making around a dollar a day in rural India often skipped cheap calorie-dense staples for variety, and many put money toward radios, weddings, and alcohol before additional food.

That is not waste. Food carries social and psychological roles beyond raw energy, and the poor trade some calories for taste, status, and pleasure the same way wealthier people do. Nutrition programs built only around more raw calories, without attention to taste, convenience, and micronutrients such as iron and iodine, therefore tend to underperform.

The authors favor interventions that raise nutrition quality directly, such as fortified foods or salt iodization, which do not require persuading someone to eat more of what they have already decided not to want.

In practice: Rural Indian households living on about a dollar a day often chose variety, sugar, tea, and festival meals over extra cheap staples when they had a little more money.

3

Cheap cures, empty clinics

The poor skip extremely cheap, high-return health measures mainly because small costs of price, distance, and inconvenience loom large against uncertain future benefits.

Immunizing a child, treating water, or sleeping under a bed net often costs under a dollar per person and ranks among the highest-return investments a poor household can make. Uptake stays chronically low even when the service is free. Information alone does not close the gap.

In Kenya, bed nets offered free versus at a small subsidized fee showed demand collapsing once any price was charged, even about sixty cents, with no evidence that paying made people value or use the net more. In Udaipur, mobile immunization camps lifted full vaccination from roughly six percent to about seventeen percent just by being reliably nearby, then to around thirty-eight percent when families received a kilogram of lentils for finishing the course.

The lentils cost far less than the savings from avoided illness, yet without them most families never completed a free life-saving series. The same pattern appears with deworming pills in Kenya: negligible costs and modest logistical hurdles sharply cut participation. A sick child today is vivid; a shot against a disease that might strike years later is not.

Health systems that assume knowledge will drive prevention miss present bias and the diffuse payoff of staying well. Small design changes—price, proximity, a modest incentive—close more of the gap than moral appeals.

In practice: Udaipur mobile camps raised full vaccination from roughly six percent to about seventeen percent by being nearby, then to around thirty-eight percent when a kilogram of lentils rewarded completion.

4

Teach to the real level

Surging enrollment failed to raise learning because curricula are pitched to an elite few; the fix is teaching each child at actual skill rather than building more schools.

By the 2000s enrollment had climbed across the developing world while literacy and numeracy barely moved. In India, Pratham found many fourth- and fifth-graders unable to read a simple sentence or do basic subtraction after years in school, because the curriculum assumed preparation only the best-ready children had.

Pratham’s Teaching at the Right Level grouped children by demonstrated skill rather than age or grade for part of the day. Randomized evaluations in Indian states showed meaningful gains in reading and arithmetic within months. A Kenya study found that simply halving class sizes did little, while hiring lower-paid contract teachers accountable for results raised test scores more—incentives and pedagogy beat raw resources.

Aspirations matter too. Parents and children often underinvest because they misjudge the payoff. In the Dominican Republic, boys shown accurate statistics on the wage returns to finishing secondary school stayed enrolled significantly longer than those given no such information; many had wrongly assumed returns were flat.

Education systems fail the poor less from missing buildings than from a mismatch between what is taught, what families believe, and what a struggling child needs on a given day.

In practice: Pratham’s skill-based grouping in Indian states produced clear reading and arithmetic gains within months, while a Kenya class-size cut alone did little.

5

Children as insurance

Poor families often have many children as a rational hedge against risk and old age rather than from ignorance of contraception, so family-planning access alone rarely cuts fertility much.

Policymakers have long treated large families as a contraception problem: open clinics, hand out pills, and family size will fall. Evidence is more mixed. In much of the developing world children function as an informal pension when savings, insurance, and state old-age support cannot be counted on.

Parents invest in more children to raise the odds that at least one can support them later. High infant mortality strengthens the logic, because some children are not expected to reach adulthood. Fertility declines have historically tracked improvements in child survival, girls’ education, and old-age security more closely than contraceptive availability by itself.

The quantity-quality tradeoff associated with Becker also appears: as the return to educating one child rises and the cost of raising many well becomes harder to bear, parents shift toward fewer, more heavily invested children. Mexico’s Progresa paid poor families to keep children in school and attend health checkups; effects on fertility timing were modest, while child health and schooling investment moved more, suggesting the channel runs through incentives for children already born.

Family decisions about children are woven into how the poor manage an uncertain future, not a simple information gap.

In practice: Progresa’s conditional cash transfers shifted child health and schooling more than they reduced the demand for children themselves.

6

Nets that snap in a drought

Informal risk-sharing among the poor is sophisticated but insufficient, and formal insurance fails to take hold even when fairly priced because trust, complexity, and liquidity block adoption.

Poor households face far more income volatility than the salaried—crop failure, illness, death of a working adult—yet almost no formal insurance. Communities built substitutes: reciprocal gifts, labor-sharing, and kin networks that spread small shocks across many households.

Those systems work reasonably for small, common shocks. They collapse under large correlated ones, such as a drought that hits an entire village at once, because everyone is short and no one has surplus to lend. Formal products have struggled to fill the hole.

Weather-indexed crop insurance piloted in India paid out automatically on measured rainfall, cutting fraud risk and paperwork. Take-up stayed stubbornly low, well under what actuarially fair pricing would predict. Farmers distrusted an unfamiliar product from a distant company, found payout rules hard to verify, and lacked spare cash at the moment of sale.

Health shocks are especially dangerous: one serious hospitalization can shove a managing family into years of poverty as medical bills meet lost income. Insurance markets for the poor are not simply waiting to be sold; they must clear deficits of trust and liquidity that actuarial fairness alone does not solve.

In practice: Indian rainfall-index crop insurance from firms such as ICICI Lombard still saw low take-up despite automatic payouts and fair pricing, blocked by trust, complexity, and cash-on-hand.

7

Why the moneylender wins

Microfinance genuinely lowers borrowing costs but cannot replace the large flexible loans needed for transformative investment, so its real achievement is smoothing consumption rather than ending poverty.

Poor households routinely borrow from local moneylenders at four to five percent per month—sixty to a hundred percent a year or more—far above formal bank rates. That is not pure exploitation. Moneylenders lack seizable collateral and court enforcement, yet hold detailed local knowledge banks do not, so they price default risk and monitoring.

Microfinance tried to undercut this by lending in small groups whose members cross-monitor each other, lowering default without collateral. An evaluation of a lender’s expansion into new Hyderabad neighborhoods found real but modest effects: existing small businesses invested a bit more and durable-goods purchases shifted, yet no surge of new firms and no dramatic jump in household income or consumption.

The deeper limit is product design. Loans are typically small, short-term, and repaid on a rigid weekly schedule—fine for restocking a shop, poor for a longer-horizon machine or land purchase that might lift a household out of poverty. Group liability, meant to curb default, also discourages ambitious but uncertain ventures because failure hurts the whole group.

Microfinance succeeds at what its design allows: cash-flow management and modest growth of existing tiny enterprises, not financing a discontinuous leap.

In practice: The Hyderabad microfinance expansion raised some small-business investment and shifted durable purchases without creating a wave of new firms or large income gains.

8

The trouble with waiting

The poor save at low rates not mainly because they have nothing to spare but because saving means resisting temptation and social claims with almost no protective tools.

Across domains, poor households often can set something aside yet struggle to shield it from impulse purchases or requests from relatives and neighbors. A simple metal lockbox given to Kenyan market vendors raised savings and health-related investment because money became harder to reach on impulse and easier to defend from others.

Rotating savings and credit associations common across Africa and Asia work partly the same way: money paid into a group pot each week leaves the saver’s control and turns a weak intention into a binding obligation. Timing can matter as much as the tool.

Kenyan farmers offered fertilizer right after harvest, when cash was in hand but planting was months away, adopted at triple the rate of farmers offered the same fertilizer at the same discount at planting time, when money was tight and immediate needs competed. The active ingredient was not price but the moment when self-control was easiest.

Poverty is often a missing infrastructure for protecting future-minded decisions. Wealthier people lean on automatic payroll deductions and retirement accounts the poor rarely have.

In practice: Offering fertilizer to Kenyan farmers immediately after harvest, at the same discount, tripled adoption versus waiting until planting time.

9

Reluctant entrepreneurs

Most businesses run by the poor are disguised unemployment kept permanently small by a return-to-capital curve that only a lucky few ever cross with a large enough infusion.

A striking share of poor households run a vegetable stall, rickshaw, or one-room shop. That sight led some, notably Yunus, to cast the poor as natural entrepreneurs held back only by capital. Evidence complicates the picture.

Randomized grants to Sri Lankan microenterprises produced large returns to capital for many male-owned firms, sometimes matching or beating formal-sector returns, yet far weaker and often negligible returns for many women-run subsistence businesses. The authors read this less as ability than as scale: a single snack vendor at a crossroads faces a limited customer base no matter the inventory, so extra money raises stock without raising profit.

Plotted across many small firms, returns follow an S-shape—low at very small scale, rising through a middle range, attractive again only at a scale most poor households never reach without a large discontinuous injection. By their own account, most would prefer a stable wage job if one existed.

Genuine growth businesses are the exception. Policy that treats every stall as a seed of expansion misreads reluctant entrepreneurship for aspiration.

In practice: Sri Lankan cash and in-kind grants raised returns sharply for many male-owned microenterprises but often did little for subsistence businesses, especially those run by women facing thin local demand.