The non-libertarian Megan stated this well a few weeks ago, when she mused that "the people I was arguing with knew their libertarian philosophy well and some econ well, but not, you know, how farming works. So they would prescribe the libertarian economist remedy of markets confident that understanding econ is sufficient to have an accurate opinion. I'd say, 'but the required assumptions simply don't hold', and get back 'but they must, because econ says'...I don't think that libertarians are impervious to evidence, but it has to be evidence in a form sanctified by academic economics. Evidence from the system itself (environment, law) was highly discounted."
Hmmm about "econ says" let's see. It is much much too charitable to conclude that what "econ says" is not just "evidence in a form sanctified by academic economics". Much of what "econ says" is theory which follows from assumptions which were explored because they had interesting implications. That is, much of economic theory is a branch of mathematics in which assumptions are made and implications derived without any consideration of evidence whatsoever.
Later, much later, economists added more assumptions to get to testable hypotheses. I am not aware of any such hypothesis which has not been rejected by the data. Then the models are tweaked to fit or "explain" the rejection of the first model. This interaction with the data would occur if economic theory were completely worthless (notice the implication does not go the other way).
Furthermore theorists look for theoretical counter examples -- and always can find them. "Economic theory" as such has no implications. I have claimed repeatedly and publicl, that for any policy you think of, I can invent an economic model such that it is optimal. No one has stumped me so far.
"Econ says" really means assumptions that are obviously false, always known to be false, introduced only to clarify thought and since proven to be false imply.
Another point. In the language of the economics profession redistribution is not inefficient. The only agreed condition under which a reform can be called inefficient is if it makes everyone worse off. The implication of the implausible assumptions is that redistribution with incentive effects is Pareto inferior than an identical redistribution without incentive effects, which is, of course, impossible.