No need for encouraging posters, an efficient tax system is enough

[this post is economics-heavy, read at your own risk]

My post on the taxation of married couples has provoked some discussion, so let me elaborate a little on the points I have made. We basically have a problem of equity and efficiency: if all taxpayers should be treated according to their ability to pay, a married couple that shares everything should be taxed as if each of them earned half the income. This is the static concept of justice, embedded in joint (also called splitting) approaches to couples’ taxation like the German.

A more dynamic concept of equity takes into account that this has consequences on the labor supply if we have a progressive tax system. The reason is that deductions as well as the lower tax rates at lower levels of income are transferable to the primary earner. When deciding if and how much to work, the secondary earner has no tax deductions and faces a starting tax rate equal to the tax rate for the last Euro of the primary earner. If you add the implicit taxation of welfare benefits, the marginal tax rate of the secondary earner sometimes increases further.

Can we show that it matters? In addition to the Swedish evidence I cited in the last post, Nada Eissa and Hilary Hoynes show in “Taxes and the labor market participation of married couples: the earned income tax credit” (2004) that the EITC for low-income households actually reduces labor market participation of families with children because of the disincentive effects is has on the secondary earner (which they equate with the woman):

[W]e find a decline in labor force participation by married women that more than offsets any rise in participation by their spouses. The reduced form models suggest that married women were 1.1 percentage points less likely to work, while married men were only 0.2 percentage points more likely to work following the cumulative EITC expansions over this period. … Married women facing the highest tax rates were 2.1 percentage points less likely to work following the expansions.

So yes, it does matter.

In general, we want taxes to generate revenue without altering economic choices by much – unless this is the intention (as with alcohol or carbon taxes). In our case here, we want to avoid altering decisions of married couples by too much when we tax them. This depends in part on the elasticity of labor supply, that is, the change in labor supply in reaction to changes in the net wage.

How then should we treat couples? In an early paper that has become the “conventional wisdom” – “Optimal tax treatment of the family: Married couples” (1983) –  Michael Boskin and Eytan Sheshinski show that the secondary earner (the one with the higher elasticity) should face lower tax rates than the primary earner. This is not only in stark contrast to the splitting approach in which the secondary earner de facto faces higher tax rates. It is also different to what a strict individual tax code would prescribe, that is, equal tax rates for both.

Patricia Apps and Ray Rees put this conventional wisdom on a more solid basis in “The taxation of couples” (2007), where they show that household productivity is an important aspect, yet empirically under-researched. I wonder, for instance, whether a potentially higher household productivity of women or their higher elasticity should, from a gender equality perspective, be treated as an empirical fact or rather as a consequence of the tax code.

Related to this last issue is Alberto Alesina, Andrea Ichino and Loukas Karabarbounis’ paper “Gender Based Taxation and the Organization of the Family” (2007). In short, they show that the bargaining position within the family is important for the labor supply elasticities. If the bargaining position favors the man, Gender Based Taxation (GBT) can increase efficiency through lower tax rates for the woman who has an endogenously higher elasticity. If the family reacts to these tax incentives by reallocating household labor, the result is a more equitable distribution of these duties which can also be more efficient. The tax code is in part responsible for the higher supply elasticity of women and there are thus two ways in which GBT can increase efficiency.

With GBT, and this is my hypothesis, the relative bargaining power will change towards a more equitable situation over time, in which case GBT becomes redundant. Until then, GBT can increase efficiency and counteract culturally shaped bargaining powers. Because let’s face it: even pragmatic liberals (in the European sense of that word) cannot deny that there is some history-dependence in the bargaining powers of men and women, for example through gender roles in society or discrimination in the workplace.

You will never get it right when taxing millions of people in millions of different circumstances. But the empirical evidence in Germany regarding family duties, female workforce participation and elasticities is clearly in favor of GBT. It will never happen, I know. But the least thing economists should do is fight for an individual tax code that stops the inefficient and inequitable splitting system we have now.

PS: There is more recent literature on the subject, for instance “Optimal Income Taxation of Couples” (2011) by Henrik Kleven, Claus Kreiner and Emmanuel Saez. But the post is too long as it is, so I stop here.

PPS: Alesina et al. wrote an FT piece on his paper, see the Economist’s View for an excerpt.