We all know what countercyclical fiscal policy is. An expansionary policy is increasing government spending, or cutting income taxes, or maybe some other tax. Oh, and its temporary, paid for by raising debt. So we have to say how the debt is financed (or maybe paid off in the future). Which means cutting future government spending, or raising future taxes of some kind. Well that makes a large number of combinations. And then there is this balanced budget fiscal expansion idea, which does not involve debt at all.
Now this multitude of possibilities would not be a problem if they all produced the same answer (the same multiplier). But they do not. I’ve talked about the various possibilities in a post before, but if you want it all done properly (as modern macro defines properly), see Denes, Eggertsson and Gilbukh (earlier version here). Now this may be fine for macroeconomists like me, because we can appear wise given this apparent confusion. But it allows others to say ‘theory is all over the place on what the multiplier is’, which is misleading. The problem is that we have not defined the policy properly.
The source of the confusion is two-fold. First, fiscal policy involves many possible instruments, and balanced budget changes in the instrument mix can have significant effects on output. Second, because a frequently discussed option involves increasing debt and never paying it off, but just paying the interest, temporary changes in one instrument can be associated with permanent changes in another. In particular, an option often considered is a temporary increase in government spending financed by a permanent increase in debt, financed by permanently higher income taxes. As I have remarked before, this policy combination is like a red rag to a bull, where the bull believes Keynesians just want to increase the size of government and raise taxes.
So how to improve things? Here is my suggestion. Let us define ‘pure’ countercyclical fiscal policy as a temporary change in some fiscal instrument, financed by changing debt, and that this debt is subsequently paid off or financed using the same instrument. So, if we raise government spending now, we payoff/finance the increase in debt by cutting government spending later (and not by raising any taxes). Now that does not make all multipliers the same, but it allows us to talk unambiguously about the government spending multiplier, or the income tax multiplier, and know exactly what we mean. Its great advantage is that, in considering the government spending multiplier for example, we do not need to address the issue of what impact any particular tax has on demand or supply at any particular time. The macroeconomics is simplified.
An alternative would be to denote one particular fiscal instrument as the residual means of finance. So, for example, it is always income taxes that pays off/finances any debt created by any particular fiscal expansion. That would also allow us to talk unambiguously about particular multipliers, but it more complicated in macroeconomic terms. We need to know what the impact of changing taxes is, even if we start by increasing government spending. Now this complication might be inevitable if policy makers have an instrument which they think of as the residual financing tool, but I do not think they do. A final problem with this approach is that it confuses two quite different issues: changing the timing of fiscal instruments, and changing their composition.
One possibly controversial aspect of my definition is that it excludes the balanced budget multiplier from the definition of countercyclical policy. But I see no problem with that. Both permanent and temporary compositional changes in the government’s fiscal mix may have important impacts on the demand for or supply of output (and inflation directly if they involve tax switches). The balanced budget multiplier is a particular type of temporary compositional change in fiscal policy.
So I like my proposed definition of pure countercyclical fiscal policy. It does not stop us analysing, or indeed recommending, more complex policy combinations, but I think any analysis would be improved by clearly thinking about such policies in two parts: a pure countercyclical policy that is just about timing, combined with some additional compositional change.