At the ZLB, reducing unemployment benefits increases total unemployment AND reduces frictional unemployment

Paul Krugman has been accused recently of being inconsistent with his own textbook on the issue of whether higher benefits for the unemployed raise or lower unemployment.  His textbook explained the consensus view that such benefits can increase 'frictional' unemployment - or the natural rate of unemployment, by reducing the pressure on the unemployed to find work.

Krugman's reply was, paraphrasing:  there are no jobs for these people to find, so not paying them benefits is just a reduction in private demand, and that aggravates the situation in the US, which is to be characterised as one of deficient demand.  Those benefits, which may be wasteful at other times, are productive because they are a way of public spending filling in the gap that low private spending has left.

But actually there's another reason that PK is not, actually, being inconsistent, that is worth teasing out.  If you buy the modern sticky price model of the liquidity trap, as exposited in papers by PK himself with Eggertson, or Woodford, and others, what the expiry of benefits for the unemployed does [on top of reducing aggregate demand] is to act like an increase in aggregate supply, by pumping up the numbers looking for work [unless all the said unemployed are completely discouraged now anyway, which one infers that they are not, not only from survey evidence, but from the fact that otherwise inflationary pressure would be much higher that it is currently].  This effect is entirely consistent with textbook PK.  We are encoding that the natural rate of unemployment falls, in line with the 1990s/2000s micro-econometric evidence on 'active labour market policies'.  But it's also consistent with blogging-PK, in the sense that it's bad.  It increases the gap between actual demand and potential output, reducing inflationary pressure further, requiring more monetary or fiscal easing to counteract.  Eggertson called things like this 'paradoxes'.  They are paradoxes because they are things that we commonly think of as good.  But actually turn out to be bad for the whole economy, and most people in it, in general equilibrium.

Chris Dillow defends PK by arguing that the effect of unemployment benefit expiry might be different in different states of the world, and that such states of the world naturally requires different models.  PK seems to accept this 'different models' defence, adding that economics is no different from, and not inferior to natural sciences in resorting to it.  But actually I think the defence is sounder than both or Krugman or Dillow make out.  The Eggertson-Krugman-Woodford, sticky-price account of the world is a single model, just one that recognises that sometimes shocks are large enough to drive the economy to the zero lower bound [which is always there, not sometimes] and sometimes they are not.

So, in brief, cutting unemployment benefits reduces aggregate demand [bad in a liquidity trap] and increases aggregate supply [also bad in a liquidity trap].

There are thought-provoking challenges to the whole sticky price macro toolkit.  [I for one read and re-read Cochrane's regular attacks, particularly those on equilibrium selection.  And I recall the Grandfather of these models, Kiyotaki, calling them 'at best, fairy stories'].  But you can't accuse them of not providing a single , perfectly consistent account of current circumstances, and of why Congress should not be cutting off benefits for the unemployed.