One of the more significant things Sir Mervyn King said today is this:

There are limits to what can be achieved via general monetary stimulus ' in any form ' on its own...As time passes, larger and larger doses of stimulus are required.

This won't be a surprise to those who doubt the efficacy of monetary policy when interest rates are around zero - though it might to those advocates of NGDP targets who sometimes give the impression that there's little difference between announcing a target and achieving it.

There's a more general point here. King is rejecting "hydraulic macroeconomics" - eombodied in Bill Phillips' Moniac - which believes that pulling policy levers has stable, predictable effects.

Of course, it's not just monetary policy which has variable effects. Fiscal multipliers also vary from time to time and place to place, depending upon whether we are at the zero bound, the extent of credit controls, the exchange rate regime and so on.

Economics is not physics or engineering: there are few if any stable, reliable parameters.

Though all this should be well known, it has implications which I don't think are properly appreciated.

For one thing, it suggests that macroeconomic stabilization is harder to achieve than supposed.If the precise efficacy of monetary policy is unknown, could we rely upon it to atabilize output if an adverse shock were to hit us this year? I doubt it.

If this is the case, it suggests economic policy should have other priorities.

The obvious one is that it should aim at boosting long-run growth through supply-side reforms. Personally, I'm sceptical of this too; long-run growth may be less malleable than supposed.

Instead, there's another alternative. If macroeconomic policy cannot be relied upon to stabilize growth, threre's a case for other means of reducing economic risk - such as a stronger welfare state or macro markets. But there's no popular taste for this. If anything, the opposite. The coalition's desire for smaller government spending and larger exports would - if it is achieved which I suspect it will only half be - tend to increase economic volatility.

However, the issue here goes beyond macroeconomics. If macroeconomic policy is not hydraulic but instead has varying efficacy, mightn't the same be true of other areas of policy, such as health and education where the effectiveness of policy levers could vary with (eg) social norms?

I fear that politicians, being selected for having an optimism about what policy can achieve, under-estimate this problem - until, of course, they have been in office for a while.