CausalPFN: Amortized Causal Effect Estimation via In-Context Learning

Jesse Cresswell (Layer 6 AI at TD) · Vahid Balazadeh (University of Toronto Vector Institute.) · Hamidreza Kamkari (Massachusetts Institute of Technology) · Valentin Thomas (Mila) · Junwei Ma · Bingru Li (Department of Computer Science, University of Toronto) · Rahul Krishnan (University of Toronto & Vector Institute)

automated causal inferenceaverage treatment effectsbayesian causal inferencecalibrated uncertainty estimatescausal effect estimation causal effectscausalpfndata-generating processesdecision-makingestimator selectionheterogeneous treatment effects observational dataprior-fitted networksraw observationstransformer architectureuplift modeling

Causal effect estimation from observational data is fundamental across various applications. However, selecting an appropriate estimator from dozens of specialized methods demands substantial manual effort and domain expertise. We present CausalPFN, a single transformer that *amortizes* this workflow: trained once on a large library of simulated data-generating processes that satisfy ignorability, it infers causal effects for new observational datasets out of the box. CausalPFN combines ideas from Bayesian causal inference with the large-scale training protocol of prior-fitted networks (PFNs), learning to map raw observations directly to causal effects without any task-specific adjustment. Our approach achieves superior average performance on heterogeneous and average treatment effect estimation benchmarks (IHDP, Lalonde, ACIC). Moreover, it shows competitive performance for real-world policy making on uplift modeling tasks. CausalPFN provides calibrated uncertainty estimates to support reliable decision-making based on Bayesian principles. This ready-to-use model requires no further training or tuning and takes a step toward automated causal inference (https://github.com/vdblm/CausalPFN/).