Does social security spending add to government borrowing? I'm prompted to ask by Daniel Furr's claim, inspired by the IFS's forecast (pdf) of rising social security spending in coming years, that any government will have to cut such spending if it is to reduce borrowing.

I ask because it's not obvious that welfare spending does add to borrowing. Think about the circular flow of income. Let's say every pound a doleite gets is spent at Tesco. A cut in the dole then means lower revenues and profits for Tesco and its suppliers - which means they pay less tax - and lower employment at such establishments, which means less income tax and more welfare spending.

It's not intuitively obvious from this that cuts in welfare spending would reduce government borrowing.

Let's formalize this intuition a little. Government borrowing, by definition, must equal the net lending of the domestic private sector and foreigners. A cut in welfare spending, then, can only reduce government borrowing if it cuts the net lending (saving) of foreigners or the private sector. How might this happen?

One obvious route is that a cut in welfare reduces spending on imports and so reduces foreigners' net lending, our current account deficit. But I suspect the marginal propensity to import from of welfare spending is small, so this is only part of the story.

Another possibility is that welfare cuts cause recipients to borrow more. From the point of view of cutting the deficit, the claim that the "bedroom tax" will plunge people into debt is not a bug but a feature. It's only if individuals get into debt that the government can get out of it. This is an accounting identity.

There are, however, other possibilities. If the welfare state gets meaner, people in work might save more for fear that the state won't support them so much if they lose their job. And it's possible - though not certain - that weaker automatic stabilizers will make businesses more fearful of economic volatility and so less inclined to invest.

In theory, then, it's not clear that benefit cuts would reduce borrowing. But what of evidence?

My chart provides some. It plots spending on social protection in 31 OECD countries for which we have data against government net lending. I'm averaging over the years 1995-2009 to smooth out cyclical fluctuations.

The chart shows zero correlation between the two. Yes, Korea had a small wlefare state and Budget surplus. But equally, Sweden's generous welfare state was accompanied by a small budget surplus in this period, whilst the meaner welfare states of Japan, Italy, the US and UK all had deficits.

Now a zero correlation means zero; there's no evidence here that a more generous welfare state must improve the government finances. But equally, this evidence is consistent with the claim that you cannot cut government borrowing by cutting welfare spending.

By all means, argue that we need welfare reform on other grounds - say to improve work incentives. But let's not kid ourselves it's a surefire way of reducing government borrowing.