Lucas critique

The Lucas critique is the argument, formulated by economist Robert Lucas in 1976, that policy evaluation based on historical statistical relationships will systematically fail because the policy itself alters the behavioral relationships that generated the data. When a policy changes, agents adjust their expectations and behavior, rendering the historical correlation between variables invalid for predicting the policy's effect. The critique is a foundational challenge to naive empiricism in economics and a direct precursor to Goodhart's law: both assert that the act of targeting a measured relationship destroys the relationship.

The Lucas critique applies wherever measurement and policy are coupled: in monetary policy, where targeting inflation expectations changes how expectations are formed; in education policy, where testing regimes change what is taught; in platform governance, where algorithmic ranking changes what content is produced. The mechanism is general: agents learn, and learning makes the system non-stationary with respect to intervention.

The Lucas critique is not merely an objection to econometric technique. It is a theorem about the impossibility of controlling complex adaptive systems through models that treat agent behavior as fixed. The system you are modeling is already modeling you.