Foreign Aid and Development

Carolina Torreblanca

University of Pennsylvania

Global Development: Intermediate Topics in Politics, Policy, and Data

PSCI 3200 - Spring 2026

Agenda

  1. Foreign Aid and Development
  2. Clemens, Radelet, Bhavnani & Bazzi (2012)

Deadlines

  • Final Project Test due today (April 15) by 11:59pm

Foreign Aid and Development

How Much Aid Flows?

Who Gives?

Who Receives?

Aid Dependence

What Goes Where?

What Is the Goal?

  • Development: Reduce global poverty
  • Humanitarian: Relief during crisis and disasters
  • Strategic: geopolitical influence, commercial access?

Does aid foster development?

Your government receives $500 million from abroad

Does it spur development, and if so, why?

Econ 101: The case for “Yes”

  • Poor countries are caught in a trap: low income → low savings → low investment → low income
  • Aid injects foreign capital and breaks the cycle
  • In capital-scarce economies, returns to investment are high

Econ 101: The case for “Maybe”

  • Any large foreign inflow appreciates the local currency
  • Imports become cheap; exports become expensive on world markets
  • Manufacturing and agriculture contract

Poli Sci 101: The case for “Yes”

  • Poor governments often cannot fund public goods: courts, schools, clinics
  • Politicians have short-term incentives, investment in public goods requires longer time horizons
  • Aid can fund investments with long time horizons that short electoral cycles make unattractive

Poli Sci 101: Against

  • A government funded by donors answers to donors, not citizens
  • If aid makes governments less reliant on taxation → less representation; citizens lose leverage over the state
  • Aid creates rents worth capturing: incentives to divert and steal

The Counterfactual

  • Question: does aid produce development in recipient countries, relative to development had they not received aid?
  • Empirically tough to answer! Aid goes to poorer, slower-growing countries and those in crisis

Macro Empirical Evidence is Mixed

  • Burnside & Dollar (2000): positive, but only with good policy
  • Rajan & Subramanian (2008): no robust effect, even with good policy
  • Hansen & Tarp (2001): positive but diminishing returns
  • Micro-level RCTs: often positive for specific, well-targeted programs

Clemens, Radelet, Bhavnani & Bazzi (2012)

Four Decades, Three Answers

  • Boone (1996): no effect on investment or growth across 96 countries
  • Burnside & Dollar (2000): aid raises growth, but only with good policy
  • Rajan & Subramanian (2008): no robust effect, even with good policy
  • Similar data, similar periods, three different conclusions

Two Fixable Problems

  • CRBB argue the divergence comes from two resolvable research design choices
  • Measurement: what counts as development aid, and over what time horizon?
  • Identification: how do you separate the effect of aid from the effect of everything that moves with aid?

Time Horizon

  • A road raises productivity for decades; food relief addresses a crisis this week
  • Testing whether all aid raises GDP this year is asking whether food relief builds roads
  • The time horizon should depend on what type of aid you are measuring

Aid Types

  • Early-impact: budget support, roads, energy, agriculture; plausible growth effect within a few years
  • Long-horizon: education, health, technical cooperation; growth effects arrive over a generation
  • No clear growth mechanism: humanitarian aid, food relief, emergency assistance

Decision 1: Lag

We receive aid in t - when should we measure development?

  • (a) Same period: potential reverse causation
  • (b) Next period: allows time for aid to work
  • (c) Much longer: plausible for infrastructure, but a lot of things change

CRBB choose (b): does Aid(\(t-1\)) predict Growth(\(t\))?

Decision 2: Aid

What type aid should spur development?

  • (a) All disbursements: food relief, technical assistance, emergency aid, budget support, infrastructure; everything in the data
  • (b) Project aid: tied to specific outputs; excludes budget support, which can also fund productive investment
  • (c) Early-impact only: budget support, roads, energy, agriculture; roughly 45-53% of total flows

CRBB choose (c)

The Causal Problem

Even with the right measure and the right lag, a problem remains.

  • Countries that receive more aid tend to be poorer and slower-growing
  • Aid and low growth travel together in the data
  • Is aid causing slow growth, or is slow growth causing more aid?

Decision 3: Confounders

Poor, conflict-prone countries attract more aid and grow more slowly.

The problem: geography, colonial history, institutions, conflict all affect both aid receipt and growth; none of these are caused by aid

FE vs. First-Differencing

  • Country FE: deviation from country long-run mean; was Nigeria above its own average when aid was higher? \[y_{it} - \bar{y}_i = \beta(\text{Aid}_{i,t-1} - \overline{\text{Aid}}_i)\]
  • First-difference: change from previous period; did Nigeria grow faster when aid increased? \[\Delta y_{it} = \beta \,\Delta\text{Aid}_{i,t-1}\]
  • FD looks at period-to-period changes

What They Find

  • Results consistent across all three reanalyzed datasets
  • +1 pp in Aid/GDP is followed on average by:
    • +0.1 to 0.2 pp in annual real GDP per capita growth
    • +0.3 to 0.5 pp in Investment/GDP
  • Positive but modest; diminishing returns set in and the effect turns negative at high aid levels

Is This Actually Causal?

  • Time-varying confounders: a crisis can simultaneously trigger more aid and hurt growth; lagging and differencing do not fully remove this
  • Mean reversion: bad growth attracts aid AND predicts recovery; aid may look effective simply because bad years do not persist

Mean Reversion

Aid and Mean Reversion

  • A country has a terrible growth year; donors respond with more aid
  • The next period, growth recovers, partly because bad years do not persist
  • A researcher observes: aid was followed by growth
  • But the recovery may have happened regardless

Measurement is Crucial!

What the Paper Argues

  • Aid causes some growth on average: modest, with wide variation across recipients
  • High aid dependence shows diminishing returns; it cannot substitute for domestic investment

Why This Matters Now

US Aid Has Collapsed

  • The US is historically the world’s largest bilateral donor
  • In early 2025, the Trump administration dismantled USAID and froze most foreign assistance
  • Dozens of countries lost their primary external funding source within weeks

Who Gets Hurt?

  • Countries where aid is 10 to 20% of GDP face acute fiscal shortfalls with no capacity to absorb them
  • Sub-Saharan Africa: highest aid dependence, weakest domestic revenue base
  • Health programs (HIV treatment, malaria, TB) face immediate disruption
  • Aid cuts reduce investment and remove shock buffers in the countries least able to absorb them