The dangers of taking theory too literally, of log(x/a+1) transforms, and place versus cows in ...
blogs.worldbank.org
When a beautiful poverty-trap model meets messy real-world data, the gap between elegant theory and empirical reality can quietly distort development policy for decades.
Poverty Trap TheoryModel MisspecificationEmpirical IdentificationDevelopment Economics
Theory Briefing
- Banerjee and Newman's 1993 poverty trap model is elegant and influential, but applying it too literally to data risks misleading conclusions.
- The log(x/a+1) transformation used to handle zero values in asset data can introduce systematic distortions that mimic theoretical predictions.
- Conflating 'place effects' with cattle ownership as a poverty proxy illustrates how measurement choices can smuggle theory assumptions into empirical tests.