Yesterday, Larry Zelenak to present the above paper, which reviews a familiar issue in light of the rise in recent decades of unmarried cohabitation. In my view, the main significance for marriage tax issues that is raised by unmarried cohabitation is the following. I view households, involving people (including couples) who in some way pool and internally allocate resources owned by different members, as an important category in distribution policy. The rise of unmarried cohabitation reduces the tax system's ability to discern "true" couples based on looking at marriage. This complicates using household information.
While the paper is basically consistent with this take (also addressed, for example, by Anne Alstott here), it places more relative emphasis than I might on the particular horizontal comparison between cohabiting couples that are basically the same, for purposes relevant to the fiscal system, except that some are married and others not. That comparison matters, on both efficiency and equity grounds, but the broader couples versus non-couples comparisons are as important or even more so.
The following is adapted from an outline that I prepared for my part of the discussion at the session yesterday, fleshed out with some of my thoughts on the particular topics.
1. The case for using household-based information
(a) Couples status versus marital status - Same point as above; while one could certainly argue that one's being married or not is relevant to how one is treated by the fiscal system, the central point for me is that understanding people's household circumstances is important.
(b) What are households and why do they matter? - In general, for people who are not in the same household, the control, use, or benefit from particular resources depends mainly on legal title. If I win the lottery, then, even if I buy the gang a round of drinks (for some reason, I have here "It's Always Sunny in Philadelphia" in the back of my mind), basically the beneficiary is me, family members aside - not, say, mere roommates even if I were still in a stage of life where I had them. Even leaving aside children, parents, and other relations, there are certain relationships, typically including but not limited to married couples, in which legal title as between particular individuals may matter less than the household's norms and rules for using the collective resources. Consider a couple with a joint bank account and/or general sharing of expenses in some way. BTW, there does NOT have to be a claim here of equal sharing - just that legal title generally matters less than internal household processes for determining how the overall resources of members should be used.
Grant this, and the fiscal system cannot meaningfully address my current economic circumstances, such as based on my income, without also considering (a) income of other household members, (b) consumption needs and productive capacities of other household members, including in non-market settings such as providing childcare, and (c) the intra-household incidence problem. On this one, suppose we had separate individual filing and that this caused high-income spouses to pay more tax while their low-income partners paid less. Would this redistribute within the household? Not necessarily - it would depend on how the household actually "works" in allocating its after-tax resources.
(c) What's the issue? - Not just separate versus joint returns, but using vs. not using, as well as different ways of using, household information. For example, having joint returns in which the rate bracket dollar amounts are doubled, relative to those on separate individual returns, can be equivalent to having separate individual returns but with income-splitting (i.e., my spouse and I are each presumed to have earned half of our combined income).
2. Limited relevance of Boris Bittker's famous "trilemma"
a. Overview - Bittker in 1977 famously set forth the "trilemma" - one can't simultaneously have progressive rates, equal taxation of same-income couples whether married or not, and marriage neutrality.
Here's a simple illustration of the trilemma. Say we have a zero rate on the first $50,000 of income, and a 50% rate above that. Ann and Bob earn $50,000 each, whereas Carol and Dave's earnings are $100,000 / zero. With separate returns and no other adjustments, Carol and Dave will pay $25,000 more in tax than Ann and Bob. With joint returns, they'll pay the same amount - just how much depends on where the joint return zero bracket amount ends - but that necessarily means that there will either be a marriage penalty to Ann and Bob, a marriage bonus to Carol and Dave, or else some combination of each.
I have great respect and admiration for Bittker, who showed his mettle once again by writing something in 1977 that people are still talking about. But I wouldn't make it as central to the analysis today as it sometimes still is.
b. Is it really a trilemma? - True, you need progressive rates for the story to get off the ground. But since we will probably decide separately whether to have progressive rates - including those, outside the income tax, that arise at the low end of the income spectrum from phasing out safety-net type benefits - it's really a dilemma, same taxation of same-income couples versus marriage neutrality, that's premised on having progressive rates. So let's consider the relevance of those two competing considerations.
c. Same taxation of same-income couples - Premised on household pooling and our difficulty both in observing the actual internal splits and in targeting tax incidence as between household members, this objective has some value, conditioned on one very important modification. There is a huge difference between a one-earner couple and a two-earner couple - say, with children, to make it especially stark - that are earning the same overall income. The one-earner couple has extra labor services available, outside the formal job market, from the one who doesn't have a market job. That individual may have decided not to work in the labor market, as opposed to lacking opportunities. Treating the two couples as the same both ignores the real difference in their resources and can lead to strongly discouraging secondary earner labor supply. Hence the powerful case for, at a minimum, secondary earner deductions or credits, childcare expense deductions or credits, etc.
d. Marriage or couples neutrality - Even without moralizing, there can be positive externalities to these relationships, e.g., the insurance benefit if both have resources and one could thus help the other upon job loss, sickness, etc. This may benefit them both, but is also a positive externality insofar as society would otherwise either have to give more $$ to the hard-luck member of the couple, or would regret that individual's bad luck.
Another, perhaps more obvious, issue in the externalities realm goes to the effect on children. While I gather the empirical literature suggests that having two parents is associated with the best outcomes for children, I don't happen to know to what extent this literature has dealt with, say, issues of correlation versus causation. E.g., suppose the benefit wasn't from having two parents, so much as from having parents who succeeded in keeping their relationship going, reflecting their underlying "types."
Note also that it is surely not true that all marriages, even with children, ought to be kept together. There are some out there as to which it's best for all concerned if they end, and overly tax-discouraging this could be bad.
e. The missing issue: secondary earner labor supply - Discussed above, but this is a really central issue that one must keep in mind and that the trilemma or dilemma leaves out. It's an issue of both efficiency and distribution, with lots of other important implications mixed in as well (e.g., concerning the broader evolution of gender roles and power relationships).
3. The rise of cohabitation outside marriage raises accuracy concerns, more (in my view) than the particular fairness concerns that the paper emphasizes
Again, a central argument of the paper is that the rise of unmarried cohabitation makes marriage penalties especially unfair, since married and unmarried cohabitants may in substance be so much alike. I might instead view the greater avoidability of marriage penalties (since one can now more easily cohabit without incurring them) as reducing fairness concerns about marriage penalties. This reflects my seeing the issue more in terms of the greater difficulty of correctly observing household status that I do indeed view as at least potentially normatively relevant.
4. The paper's case for equalizing marriage penalties and bonuses
One of the paper's main arguments is that, assuming the system otherwise remains mainly as it is today, one should try to equalize the maximum marriage penalty and marriage bonus at a particular income level. I think there's something to this, and I'd spell it out as follows: Suppose - a crucial prerequisite - that one normatively values marriage neutrality at a given income threshold. And suppose that rising departures from it have efficiency and/or equity costs that rise at more than a linear rate. E.g., just for a convenient illustration that's admittedly a bit artificial, suppose that doubling a marriage penalty or bonus makes it, in some sense, four times as bad. Then if one of the marriage penalty or bonus was $X, and the other was $3X, shifting things around so both were $2X would reduce the combined social cost of the two errors. But again, this presupposes both using marriage neutrality as one's baseline, and not having the conclusion disrupted by other considerations.
5. Marriage penalties versus bonuses towards the lower end of the income distribution
The paper shows how huge, relative to income, marriage penalties can be towards the lower end of the income distribution, by reason of the phaseout of the earned income tax credit. Surely these marriage penalties are too big, all things considered, although my preferred solution wouldn't be to shrink the EITC. The paper further argues that, given the evidence suggesting that children do better in two-parent families, we should have if anything marriage bonuses, not penalties, in this range.
While this argument has some force, under its premises, there is also another side to it. Suppose that children in two-parent households fare better than those in one-parent households. If we respond (presumably for incentive reasons) by giving the former a bonus, we have an anti-insurance system in place. That is, we reward those who are already better-off by reason of their being better-off.
The EITC, of course, already has this character insofar as someone who gets a job gets more money out of it than someone who tries but fails to get a job (assuming the former remains short of the cutoff). But while improving incentives is good, so is providing insurance rather than anti-insurance. These objectives are in conflict, necessitating tradeoffs, if we provide marriage bonuses at the low end because we believe kids do better in two-parent households.