It's really not useful to talk about tax burdens and only assess federal taxes, not state levies. People often do it, in part because it's far, far easier to calculate federal income brackets and payroll percentages rather than the shifting set of income taxes, sales taxes, and user fees that define state-level revenue collection, but you actually need to figure out a way to put them in there, or at least give a nod towards their existence. Otherwise, you have a very partial data set, which can offer up a pretty misleading impression as to the progressivity and distribution of the tax burden. For instance: Mankiw says that while the tax burden is historically low, it's lower for the poor than the rich. At the federal level, that appears to be true. But when federal tax receipts are low, states compensate with their own revenue sources. And those sources -- ranging from sin taxes to sales taxes -- tend to be regressive. So it's entirely possible that, in the aggregate, the low tax burden is more regressive than it would if simply plugged into the federal distribution. On the other hand, maybe that's not true. Point is, we need to incorporate the state data, as without it, we're missing a huge piece of the picture.