Causal Inference-Driven Digital Transformation Decision Modeling and Simulation Based on the DEVS Framework

Abstract

Digital transformation is increasingly governed as a staged, disturbance-sensitive decision process rather than a one-off technology investment. Existing simulation studies can represent temporal evolution, whereas causal estimators can correct selection bias in policy evaluation; however, the two streams are rarely integrated into a single decision laboratory. This paper develops a discrete-event system specification (DEVS)-causal framework that couples market shocks, enterprise capability accumulation, online causal learning, adaptive policy switching, and execution feedback. Six governance regimes are compared over repeated simulation replications. The Causal-Adaptive regime achieves the strongest final composite performance (85.475), the highest resilience index (13.174), and the best sustainability score (0.986). After adjustment for state-dependent treatment assignment, high-pressure intervention bundles show a negative average effect on next-window performance under inverse-probability weighting, ordinary least squares adjustment, and augmented inverse-probability weighting (AIPW). The study further includes a benchmark comparison against DEVS-only and causal-only approaches, an empirical calibration protocol based on enterprise panel and event-log variables, and an ablation test in which causal feedback is removed. The ablation indicates that causal feedback improves final performance by 2.378 points, market share by 2.151 points, customer satisfaction by 6.711 points, and resilience by 1.082 points relative to neutral adaptive feedback. The framework therefore provides a reproducible and calibration-ready route from discrete-event transformation dynamics to counterfactual governance decision maps.