Chances are Mr Osborne won't find the £10bn he needs from Iain Duncan Smith's changes

The savings will take a long time to be realised: only about 35,000 workless families have a third child each year

Iain Duncan Smith, the Secretary of State for Work and Pensions, has again proposed cutting benefits for families with three or more children where no parent is in paid work.

Of the 7.9 million families with children in the UK, about 1.2 million currently have more than two. Of these, 320,000 have no parent in paid work. If all the child benefit and child tax credits that those 320,000 families are currently receiving in respect of third and subsequent children were withdrawn immediately, this would save the Exchequer £1.4bn per year, and would of course represent a significant reduction in welfare entitlement for these families, worth £85 per week on average.

However, since the Government seems to be suggesting that changes would not affect existing claimants unless they have more children, that saving would take a long time to be fully realised: each year, only about 35,000 workless families who already have at least two children have an additional child.

Before introducing such a policy, the Government would need to consider carefully how it would be rolled out. For example, would an in-work family with three or more children who subsequently lost their job be protected? Would protection apply if a family that is out-of-work at the time of the reform finds and subsequently loses a job?

Unless careful thought is given to such issues, such a policy could have unintended consequences for the work incentives of precisely those families that the Secretary of State is seeking to encourage into employment. Any exemptions would reduce or delay the Exchequer savings even further.

Overall it seems these changes could not deliver much of the £10bn per year of welfare savings that the Chancellor wishes to find in the next Spending Review period.

James Browne is a senior research economist at the Institute for Fiscal Studies. Additional analysis by Andrew Hood.