Ah, I wrote poorly. I remain unconvinced not because I ignore the evidence, but because the papers I happen to have read so far have made flawed arguments and my own research indicates the opposite — that progressive taxation reduces inequality in the long run.
More importantly, I am indeed arguing that progressive taxes can increase long-term economic growth. As you say, taxes hinder capital formation. If taxes are too high, of course that slows growth. However, if taxes are too low, capital agglomerates too quickly. In the short term, this appears to be good for growth, as capital accrues to more skilled workers, more productive corporations, and more savvy investors. Unfortunately, if the winner-take-all phenomenon is too strong, too much capital agglomeration creates instability, which in turn reduces long-term growth. Case in point, too-big-to-fail banks.
There’s a tradeoff between efficiency and fragility. A highly-tuned race car can cover a lot of ground in the short term. For a longer race, you’d want a car less optimized, with higher tolerances and more redundancy. Progressive taxes are like buying insurance. Paying the premium is a pain, but can be a long-term gain if the price is right.
Another way to explain the effect is the (long-term) efficiency of decentralization. As capital becomes concentrated, the economy’s behavior approaches that of central planning. Taxes are a decentralizing force, and can help the economy remain a free market with distributed decision-making.
Regardless, I’ll accept that there can be a big difference between the effects on real wealth and the effects on more visible/apparent wealth. I hadn’t considered that before.