MPs have written to the Chancellor this morning calling for the introduction of a marriage tax allowance. But what if the tax and benefits system already rewards marriage?
This would challenge the idea of the 'couple penalty' - the idea that the tax and benefit system makes it more attractive for families to split up than stay together. It's not mentioned in the MPs' letter, but it's an implicit part of the debate around the issue.
In principle, it sounds like a simple thing to work out: you would just need to calculate whether the system makes people richer if they split up, or not.
The problem is that it's relatively easy to work out how taxes and benefits change if a couple split up - but harder to work out how their cost of living changes. Economies of scale kick in when you live with a partner, and you need to calculate how much cheaper bills, rent and living expenses are under one roof.
The usual way of working out this is using an equivalence scale: a ratio that estimates how much extra family members cost. Most analyses that have suggested a 'couple penalty' exists rely on these scales, and usually an OECD scale drawn up in 1982. These scales
So why not ditch the abstraction, and actually aim to understand what people in the real world need to live on based on a common framework, and then work out the relative cost of living together or apart?
We need not imagine: this June some researchers did just that (report here [pdf]
None of this necessarily makes the idea of marriage tax allowances a bad one, or undermines the aim of supporting families more generally. It does suggest extreme caution though. Policy based on a purely theoretical understanding of how couples manage money is likely to be ineffective policy. Examining the couple penalty is far from easy, but the best analysis we have suggests it doesn't exist on a systematic basis.
What's more, it suggests much of the angst about undermining marriage is misplaced. If you want the system to incentivise staying together, then relax, and rejoice: it already does.