causal

Experiments Assume No Spillover, But Markets Connect Everyone

Randomized experiments require the assumption that treating one person doesn't affect others so that untreated people serve as a control, but all economic models recognize that everyone is connected through markets and networks; this creates a deep tension because experiments miss the broader General Equilibrium effects, which is especially important for cash since every dollar given is spent somewhere, having a one-for-one accounting impact on others.

causalpending

Speaker

Paul Niehaus

Evidence Quote

“for experiments to give us the answers that we want, we have to make this assumption that when I treat you, it doesn't impact anybody else... we're not seeing those broader impacts--what we call General Equilibrium effects.”

Source

Michael Faye and Paul Niehaus on GiveDirectly— EconTalk
Created: 6/13/2026, 12:23:57 AM

Your notes

Loading notes...