EconTalk Ed Leamer on the State of Econometrics
21 snips
May 10, 2010 
Ed Leamer, an esteemed economist at UCLA, dives deep into econometrics and its intricacies. He critiques the fragility of results from model choices and stresses the importance of distinguishing between robust and weak findings. Leamer discusses the shortcomings of randomized experiments and how they may not apply broadly, while highlighting the essential role of housing in business cycles. His provocative view positions economic theory as 'useful fiction,' urging humility and transparency in empirical research.
AI Snips
Chapters
Books
Transcript
Episode notes
Randomization Helps But May Not Generalize
- Randomized experiments yield clean causal answers but may lack external validity in social settings.
- Lab-style results may not transfer because human expectations and social interactions change outcomes.
Limits Of Natural Experiments
- Natural (instrumental) experiments can mimic randomization but often raise doubts about validity and scope.
- Even valid instruments may not tell you about different policy contexts or magnitudes.
Macro Needs Experiments Hard To Find
- To infer macro causal effects like multipliers you need treated and control groups, which are hard to find.
- Historical episodes (e.g., war spending) may not map to modern policy interventions.
