This paper reconsiders the question of testing for the presence of Pareto suboptimal capital overaccumulation in overlapping generations economies. The paper allows generation‐specific technology shocks to evolve over time according to a stationary Markov chain, and assumes that an econometrician observes a finite sample of aggregate quantities. In this setting, any statistical test of the null hypothesis of capital overaccumulation with size less than one also has zero power against the alternative hypothesis of a dynamically efficient steady state. This result means that the standard assessments of capital overaccumulation based on US aggregate quantity data should be viewed as inconclusive.