I can see how "robots improve productivity", in the conventional sense of "economic output divided by input".
How they raise wages, however? They will raise *total* income, because rising productivity implies that much - but as far as I can see, approximately all of that income will go to the people owning the robots. So they will - buy more robots, I guess?
It seems to me that the NBER's research is flawed in that it focuses on certain "sectors" in isolation, without taking into account how they interact with the rest of the economy. But Steve Mnunchin's analysis is even more flawed because it's not an analysis, just a dismissal, and not, so far as we know, based on anything that could actually pass for research or rational thought of any kind.
"Labour market adjustment assitance" sounds like something that gov'ts have been talking about since at least the 1980s. It sounds nice, but in practice it means the government has to identify where (i.e. in what industries) new jobs are going to be created, and they're crap at that kind of prediction.