Taxes and Economic Growth

There is an odd gap between the debate on public finance in congress and the press on the one hand and among academic economists on the other. The debate among policy makers focuses on who pays most and on whether revenues cover expenses. The academic debate focuses on effects on money metric welfare and on economic growth. I think academic economists have a lot to learn from politicians. In particular the strange phrase money metric welfare describes welfare if one has access to and applies optimal non distortionary redistribution, that is, it has little to do with welfare in any ordinary sense of the word. Unfortuantely, conclusions from academic research shorn of context and qualifications become useful talking points in the debate. This is a problem separate from the truly pathological and highly successful. efforts of the apostate economist Arthur Laffer to sell vague arguments which never had any currency among professional economists to politicians. A key issue, as hinted above, is the effect of taxes on economic growth. There are a large number of economic models in which positive taxes on capital income are bad for growth and money metric welfare. This result has clearly influenced the policy debate. However, as usual, the strong assumptions of these models are forgotten and the conclusions are distorted in translation.

There are excellent arguments that, in the long run, capital income should not be taxed (Judd 1985 Journal of Public Economics vol 28 pp 59-83). This conclusion holds even if one considers the distribution of wealth and assumes that it is very good to take from the rich and give to the poor. It requires strong assumptions about rationality, but there is no reason to think the result is very far from correct unless people are very irrational. The argument is fairly simple and follows. First sooner or later people (or their heirs) will consume. A tax on capital income is like a tax on consumption but a constant tax on capital income is not like a constant tax on consumption. Rather if one consumes in the distant future one pays more tax than if one consumes in the near future. If one consumes immediately one pays no tax. Thus a tax on capital income is like an ever increasing tax on consumption. There is a strong result that if all consumption can be taxed at a constant rate, it should be. Thus the tax on capital income should, over time, be cut to zero.

To maintain the same revenues, the tax could be very high at first then fall quickly. With no further assumptions it is best for the tax rate to be immense (well over one hundred percent) for an instance. In effect the result is that the best way to redistribute is to seize say a third of all privately held wealth and for the government to finance itself on the income of that wealth. That’s generally called socialism. It is the optimal policy implied by the Judd model.

In the real world congress can’t do anything quickly or by surprise. With additional assumptions, Judd concludes that, even if surprise is impossible, it is better to tax capital income at a very high rate for a while then to stop taxing it. The conclusion is that, if you want to soak the rich and spread it out thin, you had better do it very quickly. The model has nothing to say about the Republican practice of rapidly increasing spending and their proposal to eliminate taxes on capital income and replace them with nothing. This is impossible. Nor does it have anything to say about the honest version of the Republican proposal which is to eliminate taxes on capital income and replace them via a delayed increase in taxes on labor income. The academic literature simply says that if you want to grab the wealth of the rich it is best to do it instantly (if possible) and, in any case, as quickly as possible.

The argument appears to be that if a capital grab and zero tax on capital income is very good, zero tax on capital income with no capital grab is good. This argument makes no sense.

There is a much broader economic literature in which it is concluded that taxes on capital income are bad for growth. Oddly in this literature it is generally assumed tht labor supply is fixed and exogenous. This does not prevent the supply siders who stress the importance of incentives to work, save and invest from indirectly appealing to this literature. One might ask what difference does this make. The answer is that it can make quite a bit of difference. A standard model changed only to allow elastic labor supply can, in some cases, imply that it is optimal to tax capital income even if the revenues are completely wasted (Pelloni and Waldmann 2000 Journal of Public Economics vol 77 pp 45-79). More generally, if elastic labor supply and endogenous human capital accumulation (learning) is considered, the optimal tax on both capital income and labor income should go to zero. That is the government should run a huge surplus then live off its wealth (site Milesi Ferretti and Roubini 1998 Journal of Public Economics vol 70 pp 237-54). This is not exactly what the supply siders are doing in practice.

As noted above, the academic literature on taxes and growth is quite different from applied discussion of tax policy. In particular, the agents in most theoretical models are immortal individuals. There is a companion literature on inheritance and dynasties, but there is no mention of the facts that workers come in two genders or that labor supply is a family decision. Decades ago, Frank Ramsey demonstrated that the most efficient taxes (ignoring distributional effects) were on sellers or buyers (it doesn’t matter) of goods and services that are supplied and demanded inelastically, that is, such that the amount supplied and demanded depends relatively little on the price. The demand for the labor of men and women appears to be roughly equally elastic, since there is no major cyclical component to the gender gap. However, it is universally agreed that women’s labor supply is much more elastic than men’s labor supply. That is many women consider whether to join the labor market, while almost all men participate. This implies that, in an efficient tax system, the tax on mens’ labor income should be much higher than the tax on womens’ labor income.

In theory a revenue neutral shift of taxes from women’s labor income to men’s labor income should cause increased total labor supply. This is generally considered efficient, because it brings the economy closer to the grab it instantly first best. Such an increase in labor supply would cause an increase in the return on capital which should, in theory, cause an increase in the rate of economic growth.

So far I have ignored distributional effects. The effect of such a shift on income distribution would be to reduce inequality. Firstly, of course, some women are single mothers and they, and their children, are overwhelmingly the people most at risk of poverty. Secondly, and counterintuitively, women’s labor income reduces the inequality of labor income of two parent families. This is counter-intuitive because the variance of women’s labor income is huge, since many women have 0 labor income. It occurs basically because the variance of the average of two random numbers is less than the average of the variance.

Now look at the actual tax code. In practice women’s labor income is taxed at a higher rate than men’s labor income for a number of reasons which are well known (I’ll write this up if anyone asks me to).

An honest supply sider would focus on eliminating such gross distortions. Of course there are very few honest supply siders. (update with permission) I found one, but then I married her so, unless she divorces me for posting this, you'll have to find one of your own.