You wouldn't guess it from some of the more simple-minded calls for cuts in welfare benefits, but changes to welfare spending involve some tricky choices. Here are a few of the trade-offs involved:

1. A big welfare state vs cyclical risk. The fact is that recessions (pdf) are unpredictable. This means that monetary and fiscal policy cannot prevent recessions because policy-makers cannot see them coming. In this context, a big welfare state acts as an automatic stabilizer, ensuring that people don't suffer big falls in income if they suffer job loss or cuts in their hours.

UK history is consistent with this. Between 1831 and 1914, when we had little welfare, the standard deviation of annual GDP growth was 2.5 percentage points. Since 1946 we've had a bigger welfare state and less volatile growth - 2.1 percentage points.

You might object that, because a big welfare state implies higher taxes and lower work incentives, this implies a trade-off between the stability and rate of growth. The evidence is ambiguous. If we look at 33 main OECD economies since 2000, there is indeed a negative correlation (of -0.24) between cash benefits as a share of GDP in 2000 and subsequent growth in GDP per head. However, this is driven by some poorer countries (such as Korea) having low benefits in 2000 but enjoying GDP convergence. If we control for the level of GDP in 2000, the correlation between benefit spending and subsequent growth is statistically insignificant.

2. Risk vs incentives. Low unemployment benefits might incentivize people to find work, but they also increase the risk of job loss. This might be inefficient. If low unemployment benefits cause people to take the first job they can get, there'll be inefficient matches between workers and vacancies. Or the fear of joblessness might encourage workers to invest more in general skills and less in job-specific ones, which could also reduce productivity.

3. Incentives to work vs incentives to work more. Tax credits which top up wages increase the incentive to move into work. For this reason, the CPAG has warned that cuts to tax credits might reduce the incentive for single parents to work. However, high withdrawal rates might of tax credits dampen incentives to move to a better job.

Where you stand on this trade-off should depend upon your view of the labour market. I suspect that if we are in an era of job (pdf) polarization, there's little point incentivizing the low-paid to get better jobs because such jobs won't be readily available.

4. High marginal withdrawal rates vs targeting. Given the existence of tax credits, we have a choice. We could target them at the low-paid, but this would mean withdrawing them rapidly as people's incomes rise. Or we could withdraw them more gently, but this would mean paying credits to people on relatively high incomes.

Reasonable people will differ on these trade-offs. And our choices will depend upon our view of the economy: how volatile is output? Are labour markets supply- or demand-constrained? How is socio-technical change affecting the availability of different types of job? And so on.

What our choices should not depend upon, however, is some arbitrary decision to cut spending based upon deficit fetishism. To do this is to place more importance upon meaningless statistics than real human lives - and this is not conservatism, but Stalinism.