This paper presents an approach to estimating work transformation matrices of large-scale concurrent engineering projects with periodically correlated work processes. A mathematical model of cooperative work is formulated on the basis of the theory of stochastic periodic autoregressive processes. The model can capture not only the cooperative processing of the development tasks with short iteration length but also long-scale effects of withholding the release of information on purpose. This strategy aims at improving the implementation of the product architecture. The solutions of the least-square estimates of the work transformation matrices and the covariances are given. Finally, the estimation accuracy is investigated based on a simulated development of a door module of a vehicle door.