On the Parameter Identifiability of Partially Observed Linear Causal Models
This work addresses the challenge of recovering causal parameters in models with latent variables, which is crucial for accurate causal inference in fields like epidemiology and economics, though it is incremental by extending prior research to more general settings.
The paper tackles the problem of parameter identifiability in partially observed linear causal models, where only some variables are observed, by providing graphical conditions for identifiability and proposing a likelihood-based estimation method that asymptotically recovers parameters up to trivial indeterminacy, with empirical validation on synthetic and real-world datasets.
Linear causal models are important tools for modeling causal dependencies and yet in practice, only a subset of the variables can be observed. In this paper, we examine the parameter identifiability of these models by investigating whether the edge coefficients can be recovered given the causal structure and partially observed data. Our setting is more general than that of prior research - we allow all variables, including both observed and latent ones, to be flexibly related, and we consider the coefficients of all edges, whereas most existing works focus only on the edges between observed variables. Theoretically, we identify three types of indeterminacy for the parameters in partially observed linear causal models. We then provide graphical conditions that are sufficient for all parameters to be identifiable and show that some of them are provably necessary. Methodologically, we propose a novel likelihood-based parameter estimation method that addresses the variance indeterminacy of latent variables in a specific way and can asymptotically recover the underlying parameters up to trivial indeterminacy. Empirical studies on both synthetic and real-world datasets validate our identifiability theory and the effectiveness of the proposed method in the finite-sample regime. Code: https://github.com/dongxinshuai/scm-identify.