Optimal Fiscal Policy in a Liquidity Trap: Comment

Robert Waldmann

dares comment on Paul Krugman’s Optimal Fiscal Policy in a Liquidity Trap. Krugman argues that, when we are in a liquidity trap, the fiscal authority should use expansionary fiscal policy to keep the unemployment rate exactly at the NAIRU. Since writing the post, Krugman has been wondering why Obama is so much less ambitious.

I try to explain after the jump.

Krugman gives one answer. The strong result requires the assumption that the government uses non-distortionary taxes. If there are deadweight losses from taxes which are convex in tax rates then one wishes to smooth the path of taxes which might imply less dramatic optimal fiscal policy.

There is something very odd in the model. Krugman seems to assume that private consumption is exogenous. He has an Euler equation in which consumption now is a function of consumption in the future, but neither is affected by Government consumption if there is other than full employment. This is odd. He clearly assumes that the government has a budget constraint so it isn’t completely clear to me how he avoids Ricardian equivalence. I think he just assumes that it is silly, but it is very odd to have an intertemporal model without intertemporal budget constraints. I think that explaining Ricardian non equivalence might help explain the appeal of relatively cautious fiscal policy.*

An easy way to get non equivalence is to have new consumers born who don’t get the benefits of current government consumption but do have to pay for it. This can mean that private consumption of the currently living is unaffected by the policy now and increased later when we get income on government bonds from the poor young (and their consumption is reduced). This seems unfair. Certainly the idea that deficits are a burden on future generations is widespread. If, pre-crisis, we had optimal intergenerational distribution, the cost of redistribution to the living from the unborn would be second order in the stimulus — that is a constant times huge squared given the size of the stimulus. I don’t see how you can have Ricardian non equivalence, intertemporal optimization and no problem with distribution (I also don’t see how he gets a multiplier of exactly 1).

In particular to have no crowding out of investment (which is not in the model) the fiscal policy has to jump back to neutral as soon as the economy gets out of the liquidity trap. I think that means paying back the debt immediately, which would be hard on the young.

However, I think the real reason is that policy makers are convinced that the pre and post liquidity trap policy is wrong wrong wrong with taxes which are too low and government spending which is too high so the public debt is too high. And, most importantly, they know there isn’t a fiscal authority — that fiscal power is divided and Obama sure doesn’t have all of it.

What that means is that however quickly the national debt should be retired, it will be retired more slowly. Krugman’s model doesn’t include investment so it doesn’t have feature that the loose fiscal and tight monetary policy gives full employment with excessive consumption and insufficient investment.

Basically, I think the issue is that
1) in the real world, debt built up during the crisis will affect consumption until it is retired which means for a long long time.
2) given Pre and post crisis fiscal policy, we are consuming too much and investing too little in normal non liquidity trap periods.

Heyyyy, wait that’s just the yakker conventional wisdom ? Can it be that it is an advance on a formal new Keynesian model ? Can it be that ordinary chit chat is like formal economic modelling except more sophisticated ? Do you really honestly still doubt that the answers to all of those questions are yes ?

update: * Ooops the model has Ricardian equivalence. Either the government runs a balanced budget while stimulating the economy (this works fine in the model) or runs up a deficit and then pays it back when the economy exits the liquidity trap. Tax timeing doesn’t matter at all, because uhm the model has Ricardian equivalence. Unanticipated government spending is still a stimulus whether or not the it is financed by a deficit (which doesn’t have any effect because of Ricardian equivalence and the use of lump sum taxes). Government spending doesn’t crowd out private spending as agents are eager to supply more at the given fixed price (which is lower than marginal cost because of imperfect competition). The present value of the stream of C+G can increase so long as the present value of the stream of leisure decreases.

Of course, in the real world, there isn’t Ricardian equivalence, so the rest of my argument is OK. It’s just a “face reality, national savings decline when the government runs a deficit; look at the data” argument not an argument required to make the model consistent.

Sorry for my confusion. Part of the problem is that Krugman wrote “The Keynesian thing about the model is the assumption that (3) holds only in expectation” when he should have written “The Keynesian thing about the model is the assumption that (5) holds only in expectation”. (3) is money demand. The right hand side of (3) depends on consumption and the nominal interest rate both of which are free variables (in particular i goes to infinity makes money demand go to zero and i goes to zero makes it go to infinity so for any M,P and C there is an i such that (3) holds). Equation (5) is aggregate supply and it holds only in expectation due to price stickiness even though prices don’t appear explicitly in (5). I think the (3) in the quote is just a typo.

In any case, I don’t think Obama or Romer or Summers believes in Ricardian equivalence so they have a choice between driving the economy to full employment without increasing the deficit (possible with a huge tax increase which is politically impossible) or crowding out investment not now, but when the economy emerges from the liquidity trap.