Liberals who support a limited public safety net are faced with a dilemma. On the one hand, they want government to fulfil the role of a provider of last resort. They envisage a situation in which people provide for the vicissitudes of life through savings, asset accumulation, private insurance, mutual assistance, the extended family, private philanthropy and an active charitable sector. The government’s job should begin when all these things have failed – but only then.

However, the crowding out effect makes this situation hard to achieve in practice. Once a government programme is in place for the case that all else fails – why try all else first? If there is a ‘free’ provider of last resort, why not take the shortcut, and make it the provider of first resort? If, for example, a government programme only kicks in once people have depleted their savings and assets, it is not irrational to respond by simply not building up savings and assets. In this way, welfare programmes initially designed for a small section of the population have a tendency of enlarging their own client base.

A possible way out of this quagmire is conditionality, meaning that government aid always comes with stringent requirements attached. The US state of Wisconsin has enacted a system has in which the equivalent of Income Support can only be received in exchange for participation in job search and work programmes. While sometimes misrepresented as a back-to-the-workhouse approach, this simply means that the daily life of a Wisconsinite welfare recipient is not that different from the daily life of their working peers.

In ‘Transforming Welfare’, my chapter in Sharper Axes, Lower Taxes, I show that this logic can be extended to cover many other government payments, both in-work and out-of-work. Meanwhile, the chapter makes the case for ending universal or near-universal benefits.

Most middle-class households would presumably not want to participate in workfare schemes, and neither would they qualify for in-work transfers any longer. This would lead to a surge in demand for private insurance products and other forms of non-state risk provisioning. Instead of trying to protect everyone from everything, the remaining welfare state could then concentrate on the genuinely deprived.

But this area requires huge organisational changes. Welfare/workfare should be devolved to local governments, largely funded through local taxation. The national government can top up local funding, but welfare provision would largely cease to be the remit of Whitehall. There would be large regional differences in welfare policies and in tax burdens. This would result in intense competition and mutual learning between local governments, under scrutiny and pressure from local taxpayers.

Combined with other measures, annual savings of over £30 billion would be possible towards the end of this parliamentary term. But the emphasis of my chapter is not on the fiscal aspect. Our present welfare system is dysfunctional, on many levels failing those it purports to protect. Reducing the system’s scope to a manageable dimension, while subjecting it to competition and transparency, is the way to break the dependency trap. That would be a good deal for the poor, and a good deal for the taxpayer.