This past Tuesday, Lane Kenworthy of the University of Arizona (Dep't of Sociology) presented some chapters from his book, Progress for the Poor. This was our yearly political scientist session - we try each year to have an accountant and a political scientist, and are also open to philosophers when someone has something suitable, in addition to inviting tax law professors (and others with public economics topics), tax practitioners, and economists.
The parts of the book that we discussed had three main theses. Here's a quick summary, including my main reactions as a reader:
Targeted versus universalistic aid to the poor - There's a long tradition arguing that the best way to help the poor is through "universal" programs such as Social Security. Wilbur Cohen, who helped design Social Security, famously said that a program for the poor will end up being a poor program. Hence, for example, in Social Security the transfers from high-earners to low-earners are smooshed in (that's a technical term) with the program's universal pension aspects, so that the redistributive part of what's going on will be less clear. Kenworthy finds the common assumptions behind universalism plausible in theory but not well-supported by cross-country data.
I'd say it's quite hard to know what to make of the cross-country data, as there as classification issues and so much is going on in each country. But I'd say that (a) the targeted versus universalist distinction is purely a matter of form, since it depends on how you define the program, (b) that's entirely consistent with the standard view, which is about optics rather than substance, and (c) the actual state of the optics seems to be that sometimes universalism fares better, other times targeting. For example, "targeted" welfare has prevailed in lieu of demogrants because it's targeted, and thus avoids the critique: "Why does Bill Gates need a demogrant?" (This critique ignores the economic equivalence of what's called a demogrant program to what's called a targeted welfare program if one treats the demogrant as taxable income and sets up the marginal rate structure the right way.) Given the context-specific optical tradeoff, it's even harder to draw general conclusions than it would otherwise be. I suspect we can't really draw confident general conclusions about how comparatively well "targeting" and "universalism" work in directing aid to the poor - it depends on the issue, the time, the society, and other aspects of the context.
The tax mix: does it matter? - The book notes standard views (from general debate, not the tax policy literature as such) regarding the relative vices and virtues of income taxes, consumption taxes, and payroll taxes. But cross-country data appears to undermine any claim of strong empirical relationships. I'd respond both that the cross-country data are hard to work with, and that all these tax bases are actually a lot more similar and overlapping than non-tax people often realize. Hence, the findings aren't a surprise to me even though I like to spend time thinking about the merits of different tax bases. A further point made by the paper, with which I certainly agree, is that, for aid to the poor, transfers / government services, etc. matter more than which tax system you use.
Are government spending and the size of government LESS different in the U.S. versus elsewhere than people commonly think? Clearly yes, if you count commonly listed tax expenditures as government spending. There are of course various issues in using the tax expenditure terminology that I have written about elsewhere. (E.g., if you treat tax-favored pensions as spending under an income tax but not a consumption tax that effectively provides the same tax treatment, you are relying on mere form to drive your adjusted "spending" measure.) But the big difference between the U.S. and, say, the countries in western Europe lies more in its having a less active "distributional" branch of government than in its having a smaller "allocative" branch.