If you want to increase taxes on zillionaires, I'm with you. But if you really want to make a dent in inequality, you should also be eager to raise taxes across the board and then spend the money on things like pre-K, health care, and so forth. That's probably where you'll get the biggest bang for the buck.
and I wrote my usual comment.
Excellent as usual. Looking at the second graph, I too thought I had a point, then I noticed that I had confused "World Bank" and "Household" (don't you hate it when that happens).
On the other hand, I think "bang for the buck" is an unfortunate metaphor. The reduction in GINI per dollar is greatest if one taxes extremely high incomes and gives the money to the poorest (as in reversing welfare reform but I will not go there). The bucks involved in "Social Security, Medicaid etc" are extremely numerous and the huge bang corresponds to a moderate bang per buck ratio.
I think the valid question is not "per the buck" but "per the unit of political capital." However, I think this means that soaking the rich is a key part of any plausible anti-inequality policy. The cost in political capital is negative, since the policy is wildly popular. I just saw a poll in which an absolute majority of GOP likely voters supported higher taxes on income over $250,000 per year. Even if one fights for higher taxes on high incomes and loses, Republicans will have fought the proposal and won. This is politically costly for them.
It is vastly easier politically to extract bucks from high income people. Also they need them less than less wealthy people. Bucks from the wealthy work just as well as funding for government programs as bucks from the non wealthy.
I actually don't understand your view (notably shared by Matt Yglesias who I also respect). I will try and fail to come up with an explanation
1) given the huge increase in government spending which we (or at least I) desire, higher taxes on moderate incomes are needed, because an attempt to raise all the money from people with incomes over 250,000 per year would imply tax rates which cause unacceptably large incentive effects (mostly of the form of making tax avoidance even more attractive). Why look at how poorly the US economy performed during the whole period when the top marginal tax rate was over 70 % (hint way better than it performed when the top marginal tax rate was below 70%). Yes, straw man argues, one might attempt to have high rates and not cause costly tax avoidance by plugging loopholes. But that never works (except for whenever it's tried). I don't find argument 1 convincing at all. I'm also fairly confident that you don't either.
2) such extremely high taxes on high incomes are politically impossible. This has the advantage of being true. But even slightly higher taxes on moderate incomes are politically impossible while moderately higher taxes on higher incomes can be enacted (see your graph). Importantly, I don't know of much of any evidence that opinions on whether taxes on high incomes are too high or too low has anything to do with how high or low those taxes are. All elected Republicans always want lower taxes on high incomes. Most other people have consistently supported higher taxes on high incomes. The marked change since 2009 has passed almost un-noticed.
I don't see how anyone can believe argument 2. It is an argument about what is politically possible based on rejecting all evidence from polling and all historical evidence about what has and hasn't been done (and therefore demonstrably doable) in the past 3 decades.
So I remain mystified. I don't find your conclusion convincing. I have no idea how you reached it. I am puzzled.