Do Taxes Matter for Long-Run Growth? Harberger's Superneutrality Conjecture


WP/95/79-EA
Do Taxes Matter for Long-Run Growth?
Harberger's Superneutrality Conjecture
by Enrique G. Mendoza, Gian Maria Milesi-Ferretti and Patrick Asea

The question of how changes in tax policy affect economic activity and
welfare is a central one in macroeconomics and public finance. Arnold
Harberger, a leading contributor to the literature on this question,
conjectured that, although the mix of direct and indirect taxes affects
investment and growth in theory, growth effects of taxation are negligible
in practice. This paper re-examines theoretically and empirically the
relations between tax policy, investment, and growth; it provides evidence
in support of Harberger's view by testing the predictions of endogenous
growth models driven by human capital accumulation. The theoretical
analysis presents a standard two-sector model of endogenous growth,
highlighting the effects of factor income and consumption taxes on growth
and investment. In particular, the analysis shows that the effects of
taxation on economic activity depend crucially on the elasticity of labor
supply and the tax treatment of the sector producing human capital.

The empirical work in this paper is based on cross-country regressions
and numerical simulations, using a new methodology for estimating aggregate
effective tax rates developed in Mendoza, Razin, and Tesar (1994). Results
using panel regressions based on five-year averages and annual data show
significant investment effects from income and consumption taxes. In
particular, higher tax rates on capital and labor income are associated with
lower private investment levels, while higher consumption taxes result in
higher private investment levels. The growth effects of taxes are, however,
small. Furthermore, they are statistically significant only in full-panel
regressions, when the time-series dimension of the data is fully considered.
These results are robust to the introduction of other growth determinants.