The article by Philip Collins on these pages last week captured very well a sophisticated scepticism about my party’s pledge to recognise marriage in the tax system. However, it also revealed one of the flaws in Blairism — its blindness to institutions and how they matter.

For a start, he writes as if there is no recognition of marriage in benefits and taxes at the moment. But this isn’t correct. Most of the benefits system does not treat us as individuals but as couples and families. It saves money by fingering any adult, notably a spouse, who should take responsibility for you if you have no income. That is why people losing jobs in the recession are finding that if their spouse has even quite modest earnings they have no entitlement to benefits.

The tax system already recognises marriage as well. Only last year the Government changed inheritance tax, for example, to give specially favourable treatment to transfers of assets to a spouse. Some readers who were persuaded by Philip Collins’s article would have been up in arms if his arguments had been applied to property that they wanted to transfer to a spouse. In fact, these tax breaks may be one reason why most middle-class, middle-aged couples end up married even if they never expected to — it is how you protect your property for your partner.

So we have a system that is all too keen to recognise marriage if you are poor, so it can cut your benefits, and if you are rich, when it protects your assets. Yet this same principle is mocked if we try to apply it to the earnings that matter to the millions of people in the middle — those people who are earning enough to keep them above benefits yet not accumulating enough wealth to be affected by capital taxes.

When you get married you promise to keep your partner for richer and poorer. It is not some eccentric right-wing cause to say this means you can be treated as one financial unit. Most OECD countries give married couples the option of either pooling their income or transferring their tax allowances for income tax purposes — France, Germany and the US certainly do. In fact, the only large OECD countries that, like us, do not recognise marriage in income tax are Turkey and Mexico. Philip Collins is advocating what one might call the Turco-Mexican model, treating everyone as atomistic individuals for income tax purposes. That is the eccentric outlier.

There is a second strand to Collins’s argument: that marriage makes no difference to our behaviour. He gets perilously close to arguing that affluent, stable couples just choose to stick a piece of paper on their relationship and call it a marriage. This is like arguing that the law of theft is completely irrelevant to whether or not we take each other’s property. It would be an odd world indeed where marriage, an institution with a sophisticated legal framework around it, as well as deep cultural roots and strong public support, did not affect our behaviour.

David Cameron has never claimed that a tax break on its own is going to transform behaviour. We understand that someone who tried to explain a deeply personal decision to marry or split up by its effect on their tax would seem very peculiar. However, tax reform as part of the wider recognition of marriage in the public realm would be a good example of a “nudge” to behaviour. Nudge, Richard Thaler and Cass Sunstein’s book calling for this approach, got a warm welcome across the political spectrum.

The main argument does not, however, depend on any nudge to behaviour. It is simply that it is unjust not to recognise the legal commitment to maintain someone else when the tax authorities are deciding on your tax liability. Don Draper and Leonard Beighton, two former Revenue officials, have measured the scale of the injustice. They calculate that the distribution of our tax burden between different types of household is broadly similar to the rest of the OECD — with one key exception. They estimate that a single-income married couple in the UK paid more than 44 per cent more tax, net of benefits and tax credits, than a single-income married couple in the average OECD country. The peculiar structure of our income tax is the reason.

There is one final challenge from the sceptics. They ask why we should not recognise all stable long-term relationships, including cohabitation. After all, our benefit system does not just take account of your partner’s income if you are married, it also counts if you are cohabiting. We do that because it would not be right if being legally married cost you benefits but cohabiting did not.

But judging if a couple is cohabiting — “snooping” — is still one of the most controversial features of our benefits system. This is a crucial clue to how a tax relief can work best. Being married or in a registered partnership is a publicly recognised fact. This does not just mean it has a much more significant impact on our behaviour than cohabiting. It is also a solid basis for tax policy without having to snoop into people’s private lives. They just have to be legally married — or indeed in a civil partnership. Going beyond that would be too intrusive.

This gives people a genuine choice. You can cohabit, as most of us do now for a stage in our lives. Then you have fewer mutual obligations to your partner and fewer privileges. Or you can choose to be married with extra obligations and some privileges too. The choice is yours.

David Willetts is Shadow Secretary of State for Universities and Skills and has responsibility for Conservative family policy

Times Recommends

• A dose of defeat is just the medicine
• Protecting ourselves from Eurocreep
• After the summer of LUV, it'll still hurt