factual

Reanalyzing seminal papers reveals fragile assumptions

By taking the three most-cited aid-growth papers (Burnside-Dollar, Rajan-Subramanian, Boone) and stripping away their differing assumptions one by one, Clemens et al. found results were highly sensitive to arbitrary data choices—such as Boone truncating all observations where aid exceeded 15% of GDP, or Rajan-Subramanian dropping available 1970s data—and restoring those observations harmonized the conflicting findings.

factualpending

Speaker

Michael Clemens

Evidence Quote

“once you get into this game of I can come up with a plausible footnote which justifies cutting whatever 10% of the data set I want, that's a route toward putting the 'con back in econometrics', as Ed Leamer once wrote.”

Source

Michael Clemens on Aid, Migration, and Poverty— EconTalk
Created: 6/13/2026, 12:10:08 AM

Your notes

Loading notes...