Weekly links October 23: macro impacts of migration on poverty, big African cities and jobs, how to read a paper, and more...

· How Nepal reduced poverty through international migration: Mahesh Shrestha has an excellent explanation of Nepal’s use of migration in In Development “In 1995, seven in 10 Nepalis lived in extreme poverty. By 2022, fewer than one in 40 did...Compared to the giants of extreme poverty reduction — China, Indonesia, Vietnam, Bangladesh, and India — Nepal’s pace is matched only by China. The extreme poverty rate in Nepal now is lower than in the considerably richer Indonesia, India, and Bangladesh. Nepal’s journey out of extreme poverty also looked very different from that of these other countries. These countries — responsible for the global decline of extreme poverty from 43% in 1990 to 10% in 2025 — did it through sustained growth, foreign direct investment, export expansion, and industrialization. Nepal had almost none of that....Nepal’s escape from extreme poverty was built not on factories or foreign investment but on the export of labor.” · African cities, urbanization, and jobs: This piece in the FT on Africa’s explosive growth in cities has sparked a bit of discussion. “African cities expanded by 15mn people annually in the past few years and are expected to grow by 19mn annually by 2050. That’s almost two Londons a year, or more than 50,000 new African city dwellers every day....A big question for Africa over the coming decades is whether rapid urbanisation will eventually lead to faster growth and higher incomes. Or alternatively, will overstretched cities with a desperate lack of infrastructure and few formal jobs become epicentres of poverty and popular unrest?”. The piece goes on to discuss how poor infrastructure, sprawling unplanned cities, and a lack of industrialization to accompany the growth have limited the jobs benefits. It gives a great example of someone who by survey definitions is working, but is not what most people would consider when they say they want a job “One young man patrols the streets with a pair of nail clippers, cutting customers’ nails for a few cents.”. My big problem with stories like this is that they don’t typically take the counterfactual of what people would be doing had they not urbanized. This UN report on African cities notes “This report presents compelling evidence - from 2,600 cities across 34 countries - that urbanisation in Africa contributes to better economic outcomes and higher standards of living. It shows that in most socio economic dimensions, Africa’s cities significantly outperform the countries in which they are located, and that the gap between the performance of African cities and the national averages is larger than in many other parts of the world. One of the most underappreciated achievements of African cities over the last 30 years has been that, despite growing by 500 million people, they have maintained their economic performance, providing several hundred million people with better jobs and improved access to services and infrastructure.” - we can both ask how can things be better, but also should ask “better compared to what”. · ETRM interviews Deniz Sanin and finds out how she got her research idea in a Starbucks. Also nice advice on reading papers “I think it’s important to read a paper from the perspective of: What’s the contribution, and what’s the gap? We have a tendency to focus just on the identification and the result. But if someone has made this contribution, what’s the next step? Sometimes it can be low-hanging fruit. Sometimes it’s just a question where someone says A, another person says B, but is it A or B, or is it maybe C?” · How well can AI do your code now? In VoxEU, Sebastian Galiani, Federico López and Raul Sosa report on a benchmark exercise of 40 econometric tasks done in Stata, R and Python with LLMs. “We built a benchmark of 40 econometric tasks covering linear regression, difference-in-differences, instrumental variables, regression discontinuity, discrete choice, survival analysis, and time series. Each task states the method and comes with a simulated dataset and a reference output computed in advance....For each task, two leading models write a script, as a chatbot that never runs it and as an agent that runs and revises it. Letting the models run their code raises the share of correct results from 79% to 97% and makes the prompt and the software matter far less. The failures that remain under the agent come from scripts that run without errors but return wrong numbers, and detecting them is still the researcher’s work.” · Subsidizing posting formal vacancies did not lead Ethiopian firms to hire more: In VoxDev, Lukas Hensel, Tsegay Tekleselassie and Marc Witte summarize an experiment they did in Ethiopia. They subsidized the cost of SMEs advertising vacancies through formal channels such as newspapers and job boards. They find firms post more white-collar vacancies as a result, but don’t increase hiring of these workers, who demand higher wages. One explanation is that inexperienced firms don’t know how expensive (in terms of wages) it is to fill these positions, learn about it by advertising, and then as a result don’t hire – but the subsidies allow them to at least test the market.