ABSTRACT
Even though the regularity approach to causality has been replaced with probabilistic theories in the philosophical debates, it still influences how causal inferences are practiced in economics. In this chapter, I argue that causal claims established on the grounds of structural equation models, theory-inspired econometric models, and cliometric research presuppose a version of the regularity view on causality. The former two methods of causal inference import (respectively, formally and informally) causal structure from economic theory and use statistics to estimate the strength and direction of relationships. Cliometric studies aim at finding constant conjunctions of events by statistical analysis of economic history. These research methods do not warrant that the causal claims established on their grounds are invariant under intervention. I distinguish between policy actions that do and do not modify the relata of causal claims and put forward the view that intervening in a former way based on evidence presupposing a version of the regularity view requires translating causal claims into the manipulationist reading. Given that this evidence does not warrant intervention invariance of causal claims, such a translation is not justified.
