As this is a follow-up to my prior post, you might want to start by reading Part 1.
The last post ended with the damsel strapped to the railroad tracks, so to speak, in the sense that I believe I had illustrated a flaw in standard tax expenditure (TE) analysis that is as basic as that in the conventional taxes versus spending distinction that David Bradford so nicely illustrated with his "weapons supplier tax credit" example.
The point I made is that looking just at the income tax, and distinguishing its "normal" features from those that are deemed to be "spending," leaves one vulnerable to framing effects, in the sense that identical policies will be described differently - by people typically convinced that the terms "taxes" and "spending" actually matter, and demonstrate something "real" (such as the size of government) - depending purely on the formalities of presentation and perhaps administration.
But there is a better way (to quote Bill McKay from The Candidate). The distinction I offered in my article drew on Richard Musgrave's classic The Theory of Public Finance (1959) to distinguish conceptually between the "distributional" and "allocative" branches of government. The former concerns who gets what, the latter how all assets in the society are used. In short, distribution and efficiency, although the latter term has to be interpreted broadly for this purpose.
When a system mainly serves one purpose and something is smuggled into it that cannot be plausibly rationalized without deploying the other purpose, you have a risk of confusion. For example, the main reason for having an income tax is distributional: we prefer apportioning tax burdens based on some measure of ability to pay to having, say, a uniform head tax. The rationale for a welfare or other transfer system is likewise primarily distributional. By contrast, the weapons supplier tax credit can only be rationalized in allocative terms - someone decided that the Pentagon should have those weapons.
Allocative versus distributional is only one example, but a very helpful one in dealing with the issues posed by TE analysis. By contrast, here's an example where something like TE analysis might help with purely allocative programs. Suppose we enacted a carbon tax, which is clearly an allocative program, but put into it a huge exception for carbon emission by homeowners. Rationale: We Love Home Ownership, it's the American Way, never mind about 2008 and all that. That, too, is an allocative rationale, albeit probably a very stupid one. But it is so clearly distinguishable from the species of allocative rationale that underlies the carbon tax that we can say: it's something separate, interposed into the carbon tax for reasons apart from trying to measure carbon emissions or the harm they cause (which of course is uniform as between carbon atoms).
By making the cut at allocative versus distributional, you can create a structure where program choices that are actually being compared to each other are all within the same analysis. So you address both the weapons supplier tax credit problem and the "distribution policy inside versus outside the income tax" set of problems that I tried to illustrate in my Part 1 post.
So how come no one is game for the enterprise that this suggests? I think I understand that a little better after yesterday's session with Len Burman than I did before. Perhaps I had been lazy about it, although I'd rather call it keeping my focus on issues where I have a comparative advantage. I have noted that step 1 of adopting a more coherent, rather than a less coherent, view of tax expenditure analysis is to make the point that it's basically nonsensical to call, say, the earned income tax credit or child tax credit "tax expenditures." Whether good or bad, they are distributional policies that can easily be done either inside or outside the income tax. They occupy the same policy space as taxing income in the first place, deciding on marginal rates, having a welfare system or not, Food Stamps, etcetera. But I hadn't gone to Step 2, though indicating that I thought it should be done, which might involve more fundamentally redrawing conceptual categories in the budget to feature allocative and distributional policy (including regulatory stuff) rather than, or more likely in addition to, "taxes" and "spending." (Counting up dollar flows is fine - it's just attributing false significance to particular subparts that we need to watch.)
An ambitious rethinking of the budget runs into serious baseline problems - though, then again, what else is new? E.g., allocative and distributional policy, or rather the outcomes produced, compared to what?
But inadequate though it is just to improve tax expenditure analysis, rather than taking on the whole shaggy beast, if one is going to go that far (in order to correct the misperceptions created by even more naive frameworks), then why not go at least one step further and recognize that the "disguised spending" frame that underlies most contemporary uses of tax expenditure analysis becomes more coherent if one interprets it as distribution versus allocation? If we are looking at provisions in the tax code - admittedly an overly narrow focus - then addressing those that advance allocative policies, rather than being aspects of adjusting overall distribution policy, appears to be exactly what people worried about the long-term fiscal situation have in mind when they say that "tax expenditures" should help pay for the needed course correction.