Dynamic scoring of fundamental tax reform: the good news and the bad news

Courtesy of Bruce Bartlett, here is a link to a pdf file of the just-released Treasury study of the dynamic growth effects of the tax reform plans reduced last year to zero acclaim by the Tax Reform Panel.

The good news (leaving aside that none of the plans has a chance of being adopted): the Panel's "Growth and Investment Tax" (GIT) ostensibly would raise national income, over the long run, by somewhere in the range from 1.4% to 4.8%. A straight-up progressive consumption tax ostensibly would do so by 1.9% to 6%. For the Simplified Income Tax (SIT), the predicted growth in national income was only 0.2% to 0.9%, but hey, that's better than nothing.

Bad news item #1: Since the plans are revenue-neutral relative to the Administration's budgetary baseline (present law minus all of the tax cut phase-outs and plus a number of unenacted Bush tax cut proposals), they might very well reduce national income relative to present law (with the phase-outs and no new tax cuts), since they result in a fiscal gap that is trillions of dollars higher.

Bad news item #2: I suspect that the models over-estimate the effects on the capital stock and economic growth of shifting from an income tax to a consumption tax. My reason for suspecting this is technical, rather than reflecting some personal hunch about saving behavior. The recent literature suggesting that income taxation and consumption taxation differ only in their treatment of the real riskless interest rate implies that the two systems are more alike than we have long thought. The real riskless rate has typically been in the 1 to 3% range, whereas the risky rate that I suspect the Treasury models use in predicting behavioral responses is much higher. To my knowledge, economic models generally have not incorporated this point as fully as perhaps they ought. The riskless rate point should also lower estimates of the deadweight loss resulting from inter-asset differences in cost recovery rate. But permanent gaps in the tax base, such as the exclusions of imputed rental income and various fringe benefits, are not directly affected by the change in thinking about timing issues.

One reason I suspect this is the magnitude of the growth rate differences attributed to the GIT versus the SIT. Even leaving aside that the former is partly an income tax while the latter is partly a consumption tax, the significance attributed to the timing point seems (admittedly at a casual glance) rather high, especially when compared with the SIT versus present law. Many economists, including for example Glenn Hubbard, have argued in print that the inter-asset distortions in the tax law are more important than the income vs. consumption tax choice, but the Treasury's dynamic analysis seems to come out the other way. Sure, theory should give way to empirics, but what we have here are estimated empirics that are themselves based on a theory.