For simple versions of the Cass-Diamond growth model where the fundamental welfare theorems of economics fail, we prove the set of minimal state space recursive equilibrium (RE) is large, with multiple RE existing in multiple subclasses of functions. This situation remains the true even when uniqueness results are known, for both OLG and in…nity-lived agent versions of the model. This implies the set of Generalized Markov equilibrium (GME) is also large. We also provide explicit iterative procedures from computing particular RE in each subclass, with some procedures globally stable from a topological persective for particular RE when their domains are restricted. All iterative methods are order stable relative to perturbations of deep parameters. Finally, we construct an simple economy where existing correspondence-based methods for computing GME fails, and propose a new method that computes the set of GME as minimal state spaceWe thank Subir Chakrabarti, Ramu Gopolan, Takashi Kamihigashi, Rajnesh Mehra, Adrian Peralta-Alva, Ed Prescott, Manuel Santos, Yiannis Vailakis, × ukasz Woźny, and especially, Robert Becker for helpful discussions, as well as seminar participants at IU-PUI, ASU, and EWGET-2012 at Exeter for their helpful comments. Manjira Datta and Kevin Re¤ett thank the Center d'Economie de la Sorbonne (CES) and the Paris School of Economics for graciously arranging their visits during the Spring terms of both 2011 and 2012.y Manjira Datta, Department of Economics, Arizona State University; Leonard J. Mirman, Department of Economics, University of Virginia; Olivier F. Morand, Department of Economics, University of Connecticut; Kevin Re¤ett, Department of Economics, Arizona State University. We are in the process of rewriting this paper, so do not recirculate, as errors, types, etc. may remain; rather write the corresponding author at Kevin.Re¤ett@asu.edu for the most recent version. 1 RE using …xed points of the expanded set of state variables. Our results point to the complicated nature of the RE approximation problem even in simple macro models.