In this paper [PDF], Staveren takes these assumptions apart:
The choice of individual utility as the unit of measurement of Pareto efficiency, however, implies that it is not resource-use that is measured for the evaluation of efficiency. Rather, the assumption is that when total utility is maximized, this can only mean that resources must have been used to their maximum efficiency. This assumption, however, is incorrect.Put plainly, we are not necessarily using our resources to create the greatest good for the greatest number. We may fail to do so because property rights are not clear and/or prices do not reflect the value of the resources.
Modern economics has recognized that preferences may include psychological desires, relying on feelings of jealousy, status, affection, etc. So, it is not resource-use that is the space in which Pareto efficiency is measured, but desire fulfillment, including desires that are unrelated to resources as well as desires that are highly resource-intensive or that are harmful for oneself but indulged in because of myopia, limited information or weakness of will. In addition, the satisfaction of some preferences generates externalities, affecting other agents' desire fulfillment.
Hence, from a resource perspective, utility maximization does not necessarily result in the most efficient use of a society's resources: various preferences actually rely on or lead to a waste of resources. So, what is actually measured by Pareto efficiency is sub-optimal total utility, not minimum resource use.
[Staveren goes on to argue that redistribution of resources (equity) can do wonders for both efficiency and/or utility maximization. Although I acknowledge the problems of resource distribution today, I do not fully agree with this heterodox perspective. Anyone interested in these issues should read the paper.]
Bottom Line: Institutions for managing resources need to reflect their full value. If not, we will not maximize their value.