Abhijit Banerjee - The Experimental Approach to Development Economics

[hymnofashes]

How do you conduct experiments when you have no control group and a virtually infinite number of confounding variables?

I know you can but it isn't trivial.

Transparency as a core value is most important, I agree. The statistics required to disaggregate all that causation will be too complicated for non-experts, and too many actors in this field are desperate, corrupt, or biased.

[MrDaniellinotte]

What the expert says is partly wrong and not new.

PARTLY WRONG: Tools were develop to support decisions about spending money - to assess project, monitor their implementation and evaluate results (see for instance CBA...).

NOT NEW: Available "tools" and development economics have been criticized long ago...

Prof. Banerjee seems also too confident about the role of data, especially when dealing with social and economic development issues.

[maradona81Mexico]

Im sorry but you just did not get the point. he's talking about

EXPERIMENTATION AND RANDOMIZATION. Thats could not be partly or complete wrong is just a NEW wave of doing economics, as he mentioned for getting better data and then evaluate better the program/aid.