> it doesn't cause retail clerks to get paid more because Wall St isn't hiring them away from their existing jobs
Nobody with an existing job actually has to switch professions for Baumol to occur. As the pay gap widens, more kids would study finance and fewer kids would consider retail an adequate career, leading to a relative shortage of retail labor, raising retail wages.
Suppose the cutoff to get into finance is at the 70th percentile of the general population and 99% of retail clerks are below the 50th percentile or otherwise have some reason not to even though those jobs already pay significantly more. How much more are they going to get paid because of that?
Or let's even suppose that the amount isn't totally inconsequential. Say they end up with an extra $1000/year. But now they're also paying $1500/year more for medicine. They're still down $500/year.
Throwing out random numbers is not a convincing argument about how things would actually work in reality.
Consider the fact that median real income tends to move up [1]. This is the metric that matters. It's the 50th percentile person, so mega rich outliers are ignored. And it weighs incomes against the CPI, which incorporates the price of medicine, construction, education, as well as consumer electronics, food, and pretty much everything people spend money on in realistic proportions. That's objective evidence that people actually get richer, even though the price of labor does tend to go up across the board, relative to goods.
> Consider the fact that median real income tends to move up [1]. This is the metric that matters.
Well, it's what matters if the CPI metric is perfect and doesn't e.g. over-weight things like food that haven't increased in cost as much as some other things.
And again, nobody is claiming that efficiency hasn't improved or that that isn't good. The issue is, if efficiency improves by 300% and then you get a 70% improvement out of it, that's bad -- people should have gotten the whole thing instead of having rent seekers capture a huge proportion of the improvement.
And for some specific subset of "efficiency improvements" the result isn't even guaranteed to be positive for the average person, so we don't need to lump them all together into an aggregate.
Your premise is that the people who work in retail have the option of studying finance or medicine. Suppose they work in retail because they scored at the 20th percentile on entrance exams and couldn't get into college.