What is a "structural" deficit?

Macroeconomists like to divide an actual deficit into two components: a cyclical deficit and a structural deficit. It reflects their distinction between automatic stabilisers and discretionary fiscal policy. I don't think that distinction is as clear and useful as macroeconomists think it is. And in ordinary language a "structural" deficit is one that is hard to get rid of. That's something quite different.

Take a simple example that macroeconomists might use to explain the distinction between a cyclical and structural deficit. The government sets government spending 'G', and a tax rate 't', on income 'Y'. The actual deficit is G-tY (forget interest payments on the debt, for now). But if we are in a recession, so Y is lower than normal, the actual deficit will be larger than if Y were at some normal level, Y*, because tax revenues will be lower than normal. The "structural" deficit is G-tY*, the cyclical deficit is t(Y*-Y), and if we add the two together we get the actual deficit.

In that simple example there seems to be an equally simple distinction between a cyclical and a structural deficit. But complicate that simple example, and the simple distinction disappears.

Suppose we go into a recession, and G increases from G* to G'. The actual deficit is then G'-tY. What is the "structural" deficit? Is it G'-tY*, or G*-tY*?

When it comes to tax revenue, we make a distinction between a fall in tax revenue that is an "automatic" result of a recession, due to falling income, and "discretionary" changes that result from a government decision to cut tax rates. We can presumably apply the same "automatic" vs "discretionary" changes in government spending and transfer payments too. A deficit that results from the operation of automatic stabilisers is "cyclical"; a deficit that results from the government's discretionary actions is "structural". So if the change from G* to G' was automatic, the structural deficit is G*-tY*, and if the change from G* to G' was discretionary, the structural deficit is G'-tY*.

But that's nonsense. In principle we could imagine two governments that do exactly the same thing, except that the first likes to specify in advance what will happen to taxes and spending in a recession, while the second likes to make those decisions when the time comes. (The first is more organised; the second likes to give the appearance of "doing something".) The first government's deficit would be all "automatic", and hence "cyclical"; the second government's deficit would be partly "discretionary", and hence "structural". Yet in both cases the deficit goes away when the recession ends.

Take another example. One government likes to have high marginal tax rates, but leaves those tax rates constant in boom or recession. A second government likes to have low marginal tax rates, but raise them in booms and cut them in recessions. In both cases the economy-wide average tax rate is the same, and varies in exactly the same way in boom and recession. In a recession, the first government's deficit is purely "cyclical", the second government's deficit is partly "structural". But again, in both cases the deficit goes away when the recession ends. The microeconomic policies of the two governments differ greatly; but in terms of macroeconomic fiscal policy they are the same.

Take a third example. Two governments temporarily increase spending (or cut taxes) in a recession by exactly the same amount. The first does it by passing a law that will expire in one year, unless renewed; the second does it by passing a law and then repealing it when the recession ends.

The distinction between automatic and discretionary fiscal policy, and hence between cyclical and structural deficits, is a political distinction, not an economic one. Macroeconomists agree that it's generally OK to run a temporary deficit in a recession. If a deficit of size D is appropriate for a recession of size R, why should we be more worried about the deficit if it took an Act of Parliament to get a deficit of size D?

What about interest on the debt? If we include it, the deficit becomes G+iB-tY. If we have a recession, and that recession causes a deficit, the debt will be higher when the recession ends, and so the future deficit will be higher even when G and tY return to normal. So even if you could make a meaningful distinction between "automatic" and "discretionary" changes in spending and revenues, and hence between "cyclical" and "structural" components of the current deficit, a current cyclical deficit will cause an increased future structural deficit.

The macroeconomists' distinction between "cyclical" and "structural" deficits doesn't make macroeconomic sense.

Here's the Globe and Mail's definition:

"Canada faces a structural deficit – a situation in which spending is permanently higher, and in this case growing faster, than revenues. It will persist even after the country escapes the recession, and is slated to hit 1 per cent of GDP, and growing, by 2013-14."

That's not a statement about the composition of today's deficit between the results automatic and discretionary changes in spending and revenues, it's a forecast about the future. It's a conditional forecast, of course. The Globe is saying that there will be a permanent deficit in future, unless....

"Mr. Page's report challenges Parliament and all Canadians to think about the sacrifices they are prepared to make in the name of deficit reduction."

Unless we make sacrifices. If something cannot go on forever, then it won't. What the Globe really means by saying the deficit is "structural" is that Canadians, and Canadian politicians, are overly optimistic about future taxes and government spending. It's a warning that our expectations of what future governments will do are inconsistent with the long run government budget constraint. I think that's actually a more useful definition of a "structural" deficit. The benchmark is not current tax rates, or current spending levels. It's current expectations about future tax and spending policies.

P.S. Of course, warning people in the middle of a recession that their expectations are overly optimistic isn't exactly helpful in ending the recession. But I'm going to leave that dilemma for now.

Comments

[David]

I would define a structural deficit as one which will occur indefinitely without a change in the level of government programs provided or in taxes. It's a small distinction, but I don't think you can start predicting the behaviour of future governments that haven't even been elected yet. Rather than start predicting how future governments will spend, which is purely speculative, it's more concrete to look at what it will cost to maintain the same level of government services.

[Nick Rowe]

David: that won't work. What's a "change"? Do you mean a change in the level of G and T, or g and t? Or what?

[Frances Woolley]

Nice post, Nick. One practical problem with implementing David's suggestion - which makes a lot of intuitive sense - is how do you apply it to health care? The Canada Health Act requires the provinces to cover all medically necessary hospital and physician services, but as technology changes, so does what is medically necessary.

[Nick Rowe]

Thanks Frances, but I'm not sure if I have a good answer to the question I posed.

Another way of posing it is this: when we project future revenues and expenditures, what exactly is it we are holding constant? To answer "no new Acts of Parliament", or something like that, seems to make the definition so arbitrary. Your example of the Canada Health Act is a case in point. If we had a different Canada Health Act, but where governments updated what was covered annually to include new technology, so we got to exactly the same level of expenditure in practice, would we want to say the structural deficit was any different? I wouldn't.

I remember Jean Chretien saying something years ago, before he was PM. Something like: "People keep saying the CPP (or something) is in a deficit, and we are headed for disaster. That's like a passenger in my car yelling 'Look out Jean! You are driving straight towards that mountain!' and I reply 'Yes, but the mountain is still one km away. When we get closer, I will turn the steering of course, so what are you yelling about?'"

Now, in Jean Chretien's example, you could perhaps have made a good case that he should have turned the CPP steering wheel sooner. But in the current case, it's not at all obvious we should take action immediately. And what does "taking action" mean, anyway?

[David Pearson]

I highly recommend this related piece on the "debt snowball" dynamic:

http://fistfulofeuros.net/afoe/economics-and-demography/the-debt-snowball-problem/

The author lays out the case for a "structural deficit": it is the deficit needed to stimulate the economy and create sufficient inflation such that the nominal interest on the debt does not, for a sustained period, exceed the growth in NGDP. The problem, of course, is that the same level of sustained deficit also increases the level of the debt, and this dilemma is the source of the "snowball" effect.

There is an underlying assumption here: that an economy needs to run a large deficit to escape deflation. One potential reason is that, in a period of deflation and falling velocity (i.e. shrinking bank credit), the only way to put money in circulation (and not just increase Excess Reserves), and therefore make credible Central Bank inflation targets, is to run large deficits. Governments, such as Greece and Ireland, try to maintain NGDP by running large deficits. However, as expectations for debt growth rise, they face a debt downgrade have to resort to fiscal austerity. However, fiscal austerity leads to lower NGDP growth, which just makes the problem worse.

The counter-argument is that the Central Bank can credibly raise inflation expectations without the fiscal authorities running large deficits. That may be true in theory, but what about in practice? Say the Fed adopts a 4% inflation target for fiscal 2011, and the Treasury proposes a balanced budget, causing the fiscal deficit to go from over 10% to zero. Could NGDP rise in this situation? I find that hard to believe. Therefore, if fiscal deficits cannot be removed without crashing NGDP, then they are structural. Nick would probably reply that they can easily be removed once the cycle turns and "organic" growth resumes; but that assumes that the NGDP cycle can turn while actors expect a massive reduction in stimulus. Another option is to "straight line" the deficit downward. Again, market expectations of austerity can affect NGDP growth, and that is precisely the problem faced by Ireland and Greece today.

So I would say the definition of a structural deficit is one that a government cannot reduce without incurring successive debt downgrades caused by expectations of falling NGDP. That is, one which the Central Bank must, therefore, eventually step in to finance. A corollary is that structural deficits exist when the habitual use of fiscal stimulus has created an economy dependent on it for NGDP growth expectations.

BTW, examples of such economies have abounded in Latin America, and now we have Greece, Ireland, Spain, potentially the UK, and eventually, also, the U.S..

[Justin Donelle]

I would say that a true structural deficit is once the government deficit is too high to still offer basic services at high tax rates. For example, if the debt levels became high enough in which an economy the government could no longer borrow(keeping in mind I'm thinking of an economy with a sound currency, gold standard for example, not ideal but an example). I think a structural deficit should be a term for which government spending is beyond the possibility of citizens to enjoy services such as health care(if it was government-run) in which all individuals would be taxed 98% for example, and the lenders would stop lending the government money, and thus be forced to engage in austerity in order to pay off the deficit. But even to that, prices would adjust to the income available and individuals would live despite the large tax rates... maybe I'm just being ridiculous in this thought. Deflation would occur with the lack of available money, or would it remain at similar levels due to tax rates? Actually, if deflation occurred, then the deficit would remain for a longer period, but prices would adjust to a good living standard? Someone answer me on that one.

The typical definition would normally be when the deficit will grow year after year if nothing is done.

[David]

By change, I mean a change in t (the tax rate) or "real G" (real government spending). In other words, a structural deficit is when growth isn't going to get us back to a balanced budget within the foreseeable future, so we have to either raise taxes or cut programs. I think thinking of cuts in terms of how citizens usually see them -- their effect on the final product -- is most useful and thus real government spending is more appropriate than actual government spending.

[Doc Merlin]

Actually one thing I would like to see is some of the recent advice given to developing countries that has been effective, being also applied to large western countries.

I think a lot of structural deficit comes from bad policies that raise the transaction costs in the labor market.

[Doc Merlin]

Speaking of transaction costs, if a country got rid of their income taxes and payroll taxes and replaced them with a VAT would they see both consumption and unemployment decrease?

[Doc Merlin]

Oh wow my bad, Nick. For some reason I got this blog confused. I was thinking of "structural unemployment." I need to pay more attention to where I am posting.

[Nick Rowe]

You had me a bit confused there Doc!