Poverty traps and the Big Push debate

Why are some economies stuck? Why do small income gains seem to evaporate without producing sustained takeoff? The poverty-trap literature provides one set of answers — that some economies have multiple equilibria, and the binding constraint is moving from a low-level equilibrium to a higher one. The Solow model has only one equilibrium per parameter set; poverty-trap models have two or more, and the policy implications are very different.

Single vs multiple equilibria

Solow's model is single-equilibrium: every economy converges to its own steady-state. Differences across countries reflect different fundamentals (savings rate, technology). There's no notion of being 'stuck' — given enough time, you reach your steady state.

Poverty-trap models introduce a non-convexity — a region of the production or savings function where the standard diminishing-returns assumption fails. This produces an unstable middle equilibrium with stable low and high equilibria on either side. Below a threshold, you converge to the low equilibrium (the trap). Above it, you converge to the high one.

Canonical poverty-trap mechanisms

1. S-shaped savings curve (nutrition-productivity trap)

Below a subsistence-level income, savings is zero — all income goes to immediate consumption (food, basic survival). Above subsistence, savings rises with income. Above some higher level, savings rate plateaus or falls (people consume more, save proportionally less).

Combined with the Solow capital-accumulation equation Δk = sy − (n+δ)k: at very low k, savings × y is below depreciation × k, capital actually decumulates, output falls, savings falls, deeper into the trap. The economy has two stable equilibria — low-level subsistence and a higher-level conventional Solow steady state.

2. Fixed costs of investment (the 'mine-or-not-mine' trap)

Many productive investments have meaningful fixed costs: a tractor, a brick-and-mortar shop, a power generator, a borehole. A subsistence farmer below a certain income can't accumulate enough to make the fixed-cost investment. They're stuck with low-productivity techniques. A farmer above the threshold can invest once and unlock higher productivity for years.

3. Coordination failures

A single firm investing in a new industry doesn't generate enough demand for inputs, doesn't attract suppliers, doesn't justify infrastructure. Several firms doing it simultaneously DO generate the input demand, the supplier base, the infrastructure case. Each firm's private return is low until the coordination point — then it's high. Without explicit coordination (Big Push), no one moves first.

4. Credit-rationing trap

Banks lend only to those with collateral and credit history. The poor have neither, so they can't borrow, so they can't accumulate productive capital, so they remain poor. Self-reinforcing.

Empirical evidence for poverty traps

The empirical record is mixed. The classical 'Big Push' theorists (Rosenstein-Rodan 1943, Murphy-Shleifer-Vishny 1989) argued the existence of traps was self-evident from cross-country data; later empirical work has been more sceptical.

• Cross-country growth regressions (Easterly 2006 'Reliving the 1950s') generally don't find robust evidence for thresholds at country level. Most countries' growth paths look more Solow-conditional-convergent than trap-bound
• Micro-level evidence is stronger. Nutrition-productivity traps (Strauss-Thomas 1998), credit-rationing traps (Banerjee-Newman 1993), poverty-and-aspiration traps (Genicot-Ray 2017) all have credible empirical support at the individual or household level
• Geographic poverty traps — Sachs et al. 2004 argued tropical disease burdens, landlockedness, and climate produce country-level traps. Influential but contested; specific mechanism (the disease channel) has strongest support
• Subsistence-agriculture traps — Carter and Barrett 2006 documented bimodal asset distributions in rural Kenya and Ethiopia consistent with micro-level poverty traps

The 'Big Push' policy

Rosenstein-Rodan (1943) argued that industrialisation of backward economies required a coordinated investment programme — a 'big push' across multiple sectors simultaneously — because no individual industry's investment was profitable on its own at the prevailing scale. Murphy, Shleifer, and Vishny (1989) formalised this with a sticky-wage model showing multiple equilibria.

The Big Push tradition is the intellectual ancestor of:

• Sachs's Millennium Villages Project (2005-2015) — coordinated interventions across health, education, agriculture, infrastructure in selected villages
• The 'poverty trap' Sachs popularised in The End of Poverty (2005)
• The Marshall Plan model of comprehensive aid coordination
• Industrial-policy programmes that simultaneously support multiple linked industries (covered in module 5)

The Banerjee-Duflo synthesis

Abhijit Banerjee and Esther Duflo (Poor Economics 2011; Nobel Prize 2019) crystallised the modern empirical-development position: a focus on specific binding constraints, identified through RCTs (next module), rather than grand theory. Traps exist; they're specific; remove the binding constraint at the binding margin and the trap unwinds.