In partial defense of Cyclically Adjusted Deficits

Tyler Cowen and Nick Rowe both make some good points on the use (or misuse) of cyclically adjusted budget deficits. However there is a danger in moving from ‘we have to be careful in using this concept’ to ‘we should not use this concept’. I think we can ask two questions here. First, when is it better to look at the cyclically adjusted deficit (let’s call it the CAD) than the actual deficit, and second when is it better to look at something else.

The main reason for the focus on CADs is quite simple. It is to stop people writing “because of the recession there is a black hole in the government’s accounts, so they will have to raise taxes or cut spending”.[1] Very crudely, there is the idea here that good policy involves balancing the actual budget. I would claim it is better policy to balance the CAD. The first switches the automatic stabilisers off, which makes the recession worse. Of course what people write would not matter if policymakers were wise to this, but unfortunately they are often not.

Nick Rowe says lets imagine two countries facing the same recession. The first has automatic stabilisers, but the second does not, so there the deficit does not increase as a result of the recession. However the second government undertakes discretionary action that mimics what happens in the first country that does have automatic stabilisers. So both countries are in exactly the same position, with the same actual deficit, but the second has a positive CAD whereas the first does not. Looking at the actual deficit tells us correctly they are doing the same thing, but looking at the CAD implies the second country is doing something wrong.

This means we need to be careful when the degree of automatic stabilisation might vary. I think this is well understood. It might be foolish, for example, to look at the US and Germany, and make inferences about how much fiscal policy is supporting activity based on comparing CADs. But for a given tax and benefit structure, it does nothing to invalidate the key point that it is better to balance the CAD than the actual deficit.

Now of course calculating the CAD is difficult, particularly when there is great uncertainty about the size of the output gap. But this calls for sensitivity analysis, not abandoning the attempt. It is true that uncertainty over the output gap means in theory that balancing the CAD could be worse than balancing the deficit, if we get the cyclical adjustment completely wrong, but I think that is very unlikely in practice. Here the UK is a great example. Last year the OBR revised down its estimate of potential output, and because the government's main fiscal rule targets the CAD, it tightened its future fiscal plans as a result. Now there is a great deal of debate about whether the OBR was right, but imagine what would be happening instead if the government was targeting the actual deficit. 

But while balancing the cyclically adjusted deficit is better than balancing the actual deficit, it is hardly an optimal policy. Here Nick Rowe’s example yields a deeper point. Presumably there is an optimal response to a recession, and that optimal response will certainly depend on many factors (like whether we are in a liquidity trap). The chances that this response is exactly the one that would occur using just the automatic stabilisers must be very small (particularly as the automatic stabilisers are largely independent of these factors). What we should care about is getting this optimal response right, and the mix between it happening automatically and manually (through discretionary action) is a secondary concern.

I also 100% agree with the point that budget deficits (cyclical or otherwise) may be very crude measures of the aggregate demand impact of fiscal policy. I never miss the opportunity of saying that fiscal policy is not one instrument, but many instruments with quite different effects. The example I like to use is a pre-announced increase in sales taxes. This may reduce the deficit when it is implemented, but before that it stimulates aggregate demand. In contrast, a pre-announced increase in income taxes will reduce demand before it comes into effect. However, before they are implemented the direct effect on the budget deficit of both is zero!

To sum up. When the idea that we should raise taxes or cut spending just because there is a recession is regarded as uneducated lunacy, and when fiscal councils routinely provide assessments of the aggregate demand impact of budget measures, and when people stop saying that Greece is not doing enough because it’s not meeting its actual deficit targets, then I think we can dispense with the cyclically adjusted budget deficit. I really do look forward to that day.

---

[1] I think the use of the analogy with a black hole in this context is a peculiarly British habit. It makes no sense as far as I can see, and for some reason irritates me immensely.