EA Forum Podcast (Curated & popular) “The joys of cash benchmarking” by ozymandias
Mar 27, 2026
A thought experiment about giving cows leads into why randomized trials are vital for reliable impact measurement. The podcast introduces cash benchmarking: giving control groups cash to see if programs actually outperform simple transfers. Several cases where cash beats interventions are discussed, plus situations where charities can still add value. A practical heuristic for choosing between cash and programs is proposed.
AI Snips
Chapters
Transcript
Episode notes
Cow Program Revealed Worse Than Cash
- Ozymandias tells a cow-vs-cash story where initial RCT showed cows improved lives, but a cash-benchmarked RCT showed cash did better.
- The story illustrates why you must compare programs to equivalent-value cash transfers, not just no-treatment controls.
Many Programs Underperform Equivalent Cash
- Cash benchmarking often reveals that many well-intentioned programs underperform equivalent cash transfers.
- Example: Huguka Ducor job training improved skills but cash caused larger short-term gains in income, assets, and well-being.
Huguka Ducor Training Beat On Skills But Lost To Cash
- The Huguka Ducor trial taught soft and technical skills and placed apprenticeships, improving some outcomes but not net wealth.
- A cash transfer of equal value produced larger increases in work hours, assets, savings, income, and subjective well-being.