Pearl's Calculus of Intervention Is Complete
This provides a foundational theoretical guarantee for causal inference methods, addressing a key problem for researchers in statistics and machine learning.
The paper proves that Pearl's do-calculus rules are complete for identifying causal effects from nonexperimental data in causal Bayesian networks, meaning any identifiable effect can be derived using these rules to express it in terms of observational quantities.
This paper is concerned with graphical criteria that can be used to solve the problem of identifying casual effects from nonexperimental data in a causal Bayesian network structure, i.e., a directed acyclic graph that represents causal relationships. We first review Pearl's work on this topic [Pearl, 1995], in which several useful graphical criteria are presented. Then we present a complete algorithm [Huang and Valtorta, 2006b] for the identifiability problem. By exploiting the completeness of this algorithm, we prove that the three basic do-calculus rules that Pearl presents are complete, in the sense that, if a causal effect is identifiable, there exists a sequence of applications of the rules of the do-calculus that transforms the causal effect formula into a formula that only includes observational quantities.