What did we learn?

Carolina Torreblanca

University of Pennsylvania

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

PSCI 3200 - Spring 2026

Logistics

Housekeeping

Today: the arc of the course
Wednesday: in classroom, open Q&A
- Final project questions
- Git, GitHub, pushing
May 10: Final project due

A quick exercise

You’ve been hired by the World Bank

They want a measure of development at the state level, within the United States

What would you measure? **Slack me the answer (last quiz!)

I. What is development?

Three countries

Saudi Arabia: rich (~$60k GDP per capita)
Cuba: poor (~$10k GDP per capita), life expectancy near US level
United States: rich, worst infant mortality in the OECD

Which country is “developed”?

How to measure development

Income
Health
Human capital
Political rights
Inequality

How do these correlate? Something else we misssed?

The choice is normative

If every axis covaried perfectly, development would be a single ladder

Development is multidimensional, picking a measure is a normative call

The perils of measurement

Three choices baked into any development index

What dimensions count
How they are weighted
Whether it measures levels or changes

Not the historical norm

Per-capita income fluctuated in a narrow band for most of recorded history

The long view

A working definition

Development is sustained, compounding improvement in material conditions

Sustained ?
Compounding ?
Material ?

II. What have we learned about development?

Three threads kept reappearing

We have a definition. What did the course teach about achieving it?

Accountability: who answers to whom
State capacity: who can deliver
Distribution: who receives, who pays

Democracy

Theory: informed citizens hold governments accountable

Evidence: report cards on Ugandan clinics raised utilization, cut child mortality (Björkman and Svensson 2009)

Catch: this works only where elections can punish (Ferraz and Finan 2011; Chong et al. 2015)

Information binds contingently

Representation

Theory: identities of politicians shape priorities

Evidence: reserved seats for women in India shifted spending to water, child health, schools (Chattopadhyay and Duflo 2004)

Catch: representation shits policy only when interests differ / not crystalized

Security

Promise: security is the foundation of development

But also: development sustains security. The arrow runs both ways.

Evidence: less productive years raise civil war risk (Miguel et al. 2004); Ugandan child soldiers earned less for decades after the war (Blattman and Annan 2010)

Insecurity blocks development. Underdevelopment fuels insecurity.

Crime

Promise: democratic accountability should improve policing

Catch: policing shapes who participates. Punitive contact produces custodial citizenship (Weaver and Lerman 2010)

Puzzle: cross-nationally, crime victims are more engaged, not less (Bateson 2012)

Who and how the state polices is cause and consequence of participation

Migration

Money. Education. Conflict. Opportunity. All push people to move.

How does movement reshape sending and receiving countries?

Sending: lose workers, gain remittances. Filipino remittance windfalls funded schooling and microenterprise (Yang 2008)

Receiving: gain labor. What rights for those who arrived?

Service Delivery

The biggest bang per dollar. Also the slowest.

Health: north of China’s Huai River, coal subsidies cut life expectancy ~5 years (Chen et al. 2013)

Education: Indonesian school construction raised lifetime earnings, where labor markets absorbed graduates (Duflo 2001)

Big returns. Short political horizons. Who has incentives to provide?

Climate

A global collective action problem with unequal stakes

Evidence: 1°C of warming cuts poor-country growth ~1pp. Rich countries are less affected (Dell, Jones, and Olken 2012)

Mitigation: rich countries emitted the carbon. Few electoral pressures to cut

Adaptation: the bill lands on poor countries with the least capacity to pay

The puzzle: how do you solve a global issue without supragovernment authority?

Aid

Promise: aid moves resources to where the need is greatest

But aid is non-tax revenue. It can build state capacity, or substitute for it

And: it shifts whom the state must answer to: from citizens (who pay taxes) to donors (who set priorities)

Evidence: aid concentrates where politically convenient for donors, bypassing the poorest within recipients (Briggs 2016)

Aid funds development. It also reshapes accountability.

Suppose we know what works

Security. Health. Education. Clean energy. Accountability. Representation.

All can be good for development.

But funds are limited. Capacity is finite.

Where do we invest first?

The triage problem

Health: fast, large gains per dollar
Education: slow, compounding, contingent on labor markets
Violence reduction: a precondition for anything else
Anti-corruption: hard to engineer from outside
Redistribution: politically explosive, but tractable

And the externalities?

Programs rarely affect only their targets

Cash transfers spill into local prices and demand
Conflict spreads across borders and generations
Climate adaptation in one place lowers risk elsewhere
Schooling has labor market effects beyond the educated

III. How do we evaluate development policy?

How do we evaluate policy?

A counterfactual question.

We want to know: what would have happened without the policy?

Constrained comparison

*Randomization (RCTs)
Natural experiments and instruments
*Within comparisons (FEs)
*Difference-in-differences

Each constrains the comparison differently and rests different assumptions

A claim in the wild: “Aid recipients are still poor, so aid is wasted”

Aid flows to poor countries because they are poor

Comparing recipients to non-recipients estimates the selection rule, not the treatment

Counterfactual: what would recipients look like with substantially less aid?

V. What do we do?

What we know

High confidence

Direct health interventions (vaccines, malaria, deworming) are cost-effective almost everywhere
Cash transfers outperform paternalistic alternatives, with no labor supply hit
Schooling expansion has positive returns, moderated by labor market absorption

What we don’t know

Low confidence

How to cause institutional change from outside
How to accelerate democratic consolidation
Why industrial policy worked in a handful of countries and not others

Effectively unknown

Climate adaptation at scale
Reconstructing legitimate states after conflict

Where should we invest?

The defensible answer: human capital
The harder answer: state capacity, which is difficult to build from outside
The urgent answer: climate adaptation and CC mitigation