I woke up this morning with a slight headache, so I took a paracetamol and now feel fine. But if one paracetamol makes me feel OK then surely a hundred will make me feel fantastic.

You might spot an error in this thinking. It is, though, the same mistake that some critics of me (and Tim Harford and Martin Wolf and Ryan Bourne) are making when they question our hostility towards the proposed rise in the minimum wage.

I'll grant that most studies (pdf) of the impact of the UK minimum wage have found no significant impact upon employment - although (pdf) there (pdf) are exceptions to this (pdf). But there's a reason for this; the NMW has been set at a low level. The Low Pay Commission has helped ensure that the NMW has been administered in a safeish dose. But it does not follow that a higher NMW would be safe. Just as paracetamol is safe in a low dose but dangerous at a high one, so too are minimum wages. The fact that most - not all but most - international studies show that minimum wages do reduce demand for labour shows us that, where they are set carelessly, they can do harm.

Let's consider three possible ways in which this view might be wrong.

First, employers might simply accept a squeeze on profits. However, which this might be tolerable for small wage increases, it'll not be possible for bigger rises which might make the difference between profit and loss. Firms which struggled by on a modest NMW might shut if there's a higher one.

Secondly, firms might respond to higher wages by increasing efficiency; there's some evidence that this has happened, and it is possible because there is a long tail (pdf) of badly managed firms where efficiency gains might be made. Three things, though, make me question how far this is possible:

- Whilst it might be possible for firms to raise productivity by one or two per cent, the big gains necessary to offset a 13% rise in the NMW (on the OBR's estimate) might be harder to find.

- Efficiency in this context often means work intensification: chambermaids cleaning more hotel rooms, care workers seeing more clients, and so on. This means workers might pay for higher wages with more stress.

- If productivity is to rise without job cuts, then output must increase: this follows from the definition of productivity. But how will this extra output be sold? One possibility is that aggregate demand rises. But, in tightening fiscal policy, Mr Osborne has done nothing to permit this. The other possibility is that firms will cut prices. But the OBR expects the opposite to happen.

Another way in which a higher minimum wage might be consistent with rising employment is if there is monopsony - if workers are paid less than their marginal product.

Again, I'll concede that this might be true (pdf) in some cases. However, a much better way to tackle the problem of monopsony is to increase workers' bargaining power - through stronger trades unions, out-of-work benefits which are high enough to allow people to reject rip-off job offers or through a full employment policy. These allow wages to rise where workers are under-paid without forcing wages above marginal product, which would destroy jobs. However, Mr Osborne has rejected these alternatives in favour of the blunt instrument of a law which bears equally upon monopsony and upon competitive labour markets. This is surely sub-optimal.

My point here is a simple one. The economy doesn't operate exactly as the textbooks say. Wages aren't exactly equal to marginal product and firms don't operate at maximal efficiency. Instead, as Adam Smith said, there is a great deal of ruin in a nation. This ruin provides room for minimum wages to rise a little without hurting jobs. But is it really plausible that there is so much ruin as to permit a big rise without adverse effects?

Mr Osborne has given us no reason to think this. The kindest thing one can say about his plan is that it is a bold experiment. There are less kind things.