'Growth is being depressed by the euro area crisis'. However, unemployment was rising before the latest leg of that crisis. Also, the consensus expects growth to be weaker this year because of lower consumer and capital spending, not just lower exports. The consensus now forecasts a drop of 1% in consumer spending this year against the OBR's forecast of a 0.6% rise and a fall of 0.9% in investment against an OBR forecast of 2.3% growth. These two differences take 1.5 percentage points off GDP, whilst the consensus' lower export forecast (5.3% growth vs. 7.9%) takes only 0.75 percentage points off growth. This tells us that something has gone wrong with the domestic economy, and that the euro area isn't entirely to blame.

'Gilt yields are low so the market has faith in our plans.' However, spreads between 10 year gilts and their US and German counterparts, at 0.24 and 0.28 percentage points respectively, are within a standard deviation of their post-2001 averages, of 0.29 and 0.53 percentage points. This means it's hard to infer that fiscal policy has led to significantly lower yields. And even if it had, I'm not sure this is a wholly good thing. Yields can be low because markets are pessimistic about growth, not just because they are optimistic about creditworthiness.

'Things would have been even worse under Labour's plan.' This is just unobservable. It also runs into the problem that there isn't a massive difference between the two policies. As Fraser says, Osborne's plan is an 'only-slightly-modified version of Darling's deficit reduction plan.' But it's hard to have it both ways - to claim both that there's little difference between the two policies and that Osborne's plan has made a material difference in preserving market confidence. (Of course, mutatis mutandis this is an embarrassment for Labour as well).

I suspect that arguments like these are what Popper called 'immunizing strategies' - they are ways of protecting yourself from admitting you were wrong.

This is not (just) a partisan point; all politicians do something similar. Instead, my point is that policy-making is a semi-science. Like science, it conducts experiments - in the sense of doing things the result of which are unknown. Unlike science, it just doesn't, and cannot, learn from such experiments.