A Markov Chain Framework for Assessing Owner Payment Reliability across Multiple Construction Projects

Abstract Timely payments from project owners are critical to contractor cash flow and project continuity, yet quantitative methods for assessing owner payment reliability across multiple projects remain limited. This study develops an Owner Payment Reliability Model (OPRM) based on absorbing Markov chain theory. The model represents payment claims progressing through discrete states—from submission to payment or dispute—and estimates age-dependent payment probabilities. Three formulations capture different temporal dynamics: Stationary, Nonstationary Claim-Phased (NStat-CP), and Nonstationary Interval-Phased (NStat-IP). Bayesian posterior means (using Jeffreys and Uniform priors) are used to integrate evidence across an owner’s project portfolio. The models are calibrated using 34 projects from 11 owners and evaluated using the Akaike information criterion (AIC) and Bayesian information criterion (BIC), which balance model fit against complexity to penalize overfitting. Results show that while the NStat-CP model achieves the best statistical fit, the Stationary model offers a favorable balance of accuracy, interpretability, and computational efficiency. For contractors, the framework supports pre-tender owner comparison, bid prioritization, and cash-flow forecasting by providing a continuous, project-independent measure of payment reliability. This study contributes a unified probabilistic framework for aggregating and normalizing owner payment behavior across multiple projects and demonstrates the practical trade-offs between stationary and nonstationary formulations.