Might fiscal austerity have even larger costs than generally supposed? I ask because of the interaction between fiscal policy and financial regulation.

Ed Balls says: 'An exclusive focus on financial stability could bring a dangerous risk aversion and stifle the real economy.' This expresses a well-known trade-off; we can prevent financial crises by stopping banks lending, but this also stops growth. Given that so much productivity and innovation comes from new companies and market entrants which are often reliant upon bank finance, this trade-off is acute.

So, how can we alleviate it? One way is to reduce the costs of financial crises when they happen, for example by splitting investment and retail banking so that troubled banks can be resolved, or by requiring banks to hold contingent convertible bonds which allows them to be automatically recapitalized in a crisis. In fairness, much of the Vickers report focused upon these issues.

If the costs of crises were low, there'd be less need for tight regulation of banks to prevent crises. Banks would thus be freer to lend, which would help the economy to grow.

And here's my concern. One way of reducing the cost of crises is to have an active fiscal policy. If governments can increase borrowing greatly, the economic costs of crises are mitigated.

However, two things might prevent governments borrowing in future. One is the euro area's fiscal compact, which limits structural deficits to 0.5% of GDP. The other - more relevant for the UK - is deficit fetishism. A financial crisis, pretty much by definition, will automatically increase the government deficit. This is because a fall in lending causes some people to become forced savers, and if net private sector saving rises, government borrowing will rise pretty much by identity. A government which takes fright at a big deficit will thus undertake less discretionary fiscal easing than it should. Which means that financial crises hit the real economy harder.

And here's the problem. If future fiscal policy is irrationally constrained so that the costs of financial crises are greater, then there is more need to prevent them. But this means tighter financial regulation and slower growth now.

I'm making two points here, one 'leftist' and one 'rightist'.

The leftist point is that the costs of anti-Keynesian fiscal policy are greater than thought. This is because such policy raises the cost of future financial crises and so requires tighter financial regulation now, and hence slower growth.

The rightist point is that this implies that the conventional way of thinking about regulation is mistaken. Very often, such thinking asks: what can rational governments do to constrain the irrational private sector? But this misses the point that governments can be irrational too - not just now but in the future. And if we consider this - as we must - debate about regulation becomes more complex.