Are deficits good or bad? That depends. Sometimes they are good, and sometimes they are bad.
But even when deficits are good, don't use bad arguments to defend them.
It really annoys me.
First as an economist, because it's bad economics.
Second, as a (Canadian, but whatever) citizen. Because we want governments to run deficits when it's good to run deficits, and not run deficits when it's bad to run deficits. And if people use bad arguments to justify deficits, even at times when deficits might be a good thing, we might continue to get deficits even when deficits are a bad thing, if people continue to believe those bad arguments. And I don't want Canadian policy on deficits to get screwed up by bad arguments coming in from south of the border or anywhere else. Especially since the Canadian (Federal) policy on deficits has been reasonably good (in the sense of "not obviously horribly wrong in a big way") since about 1996. Martin/Chretien ran surpluses in the good times, and Flaherty/Harper ran deficits in the recession, and are slowly eliminating the deficit as the economy recovers.
Here's a list of arguments for deficits:
0. "There is no burden on future generations because the extra taxes to service the debt will be paid to those same future generations that inherit (sic) the bonds." That is a really bad argument. It's just plain wrong, unless you believe in Barro-Ricardian Equivalence, which the people making that argument do not believe. Because if they did believe in Barro-Ricardian Equivalence, they would not be saying that increased deficits are a good thing to stimulate demand today (they might argue for balanced budget increases in government spending). So be honest, and don't use that argument.
1. "Yes, future taxes will be a burden on future generations, but monetary policy won't work at the zero lower bound, and we need fiscal deficits to increase demand now, and the benefits exceed the costs." I happen to disagree with the assumption that monetary policy can't work at the ZLB. But I might be wrong. And if it's really important to get out of a recession (which it is) there is something to be said for wasting your grandkids' money to buy suspenders even if you think your belt will keep your pants up.
2. "Yes, future taxes will be a burden on future generations, but the roads and schools etc. we are borrowing to build will be an even bigger benefit to those future generations". Good argument.
3. "Yes, future taxes would be a burden to future generations, but all we are doing is building the same roads and schools a couple of years earlier than we would have done, and doing it when roads and schools are cheap to build, and real interest rates are very low or even negative, so there's a chance that future taxes might even go down if we run a deficit today and pay it all back by building fewer roads and schools in a couple of years." Another good argument.
4. "Yes, future taxes would be a burden to future generations, but the rate of interest on government bonds will on average remain below the growth rate of the economy forever, so we can increase the debt, rollover the debt plus interest forever, without ever having to increase taxes on future generations, who will actually be better off because they will get a higher rate of return on their savings." This argument is valid. But I admit it does scare me, because we don't know for sure what future interest rates and growth rates will be. And because if we overdo it and issue too many bonds, the rate of interest will rise higher than the growth rate. And it would be hard to explain to the public why Ponzi schemes can be stable. But if that's your reason for believing deficits are a good thing, there's nothing for it but to try to explain Samuelson 1958 to the general public, and convince them your assumption is correct!
5. "Yes, future taxes will be a burden on future generations, but future generations will be richer than we are, and it's OK to transfer wealth from richer to poorer". OK. Maybe.
6. "Running deficits in bad times and surpluses in good times but balanced budgets on average doesn't mean higher taxes on future generations, and it's better than having government spending procylical and tax rates countercyclical". Yep, because government spending has diminishing marginal benefits, and tax rates have increasing marginal (deadweight) costs, so we want to keep them smooth over time. Standard micro public finance. Quite apart from any Keynesian macro automatic stabiliser arguments.
There are probably other good arguments I've forgotten. But don't use any of that "we owe it to ourselves" nonsense.
Worthwhile Canadian Initiative 
How is that the counter examples of debt can be a burden has become there will be a burden (unless something mitigates it)?
Roger: these are my counterexamples to my counterexample. Maybe you have some more?
But look, it’s really really simple. If you tax someone, they are worse off. Unless there’s some indirect benefit.
Is the need for Ricardian Equivalence in 0 contingent on the point about inheritance?
Because if you just sell the bond when you’re old and then spend the proceeds before you die we can dispense with inter-generational altruism and any reliance on strict Ricardian equivalence.
Or am I missing something?
Well I’m thinking more in the line of the original Krugman argument and the milk critique(or Rowe critique). We can affect unborn people through a systematic redistribution scheme, but not with a one time sting right? Thus in the Krugman sense of unborn generations any temporary rise of debt will be a wash (someone will have payed but not the unborn ones), right? I just read all the posts relating to the topic so I can easily be completely confused.
PS: Just in case I still have your attention before the tsunami of comments arrives: I=S it is an accounting identity because undesired savings are an investment as well, but if the stock is perishable (intra or even inter period) the firm just loses all ( or part) of the unwanted investment and thus S>I. How tired I am?
“There is no burden on future generations because the extra taxes to service the debt will be paid to those same future generations that inherit (sic) the bonds”
What is your assumption today about those future taxes? I can somehow understand why you want to call debt as burden today given the structure of today’s tax system, i.e. mostly labour and consumption. But if you assume that the tax system can change in the future (and it really can and even most likely will in whatever ways) how can you insist to call it burden? Example: what if in the future there will be a 100% tax on any capital gains on government bonds. Equivalently, the whole yield curve falls pretty much to zero. Equivalently central banks does QEUnlimited.
Nick:
How about this argument: “Yes, monetary policy is not limited at the zero bound, but due to political and institutional reasons it is. Therefore, running a budget deficit at the zero bound is useful because it creates additional safe assets that can help reduce the excess demand for such assets.”
Or this not so good one:
“Yes, monetary policy is not limited at the zero bound, but due to political and institutional reasons it is. Therefore, running a budget deficit at the zero bound is useful because it can create the expectation of a permanent increase in the future monetary base to pay off the debt.”
Daniel: I’m not sure if I’m misunderstanding you, or you are misunderstanding me.
If Ricardian Equivalence is true, then there is no net burden of the debt, because we will bequeath (as a freebie) the bonds to our kids, and they to their kids, to exactly offset the burden. Future generations will then inherit (in the true sense, as opposed to “buy”) the bonds, as well as the tax burden.
Roger: “We can affect unborn people through a systematic redistribution scheme, but not with a one time sting right?”
Not right. Suppose we borrow to make transfers to people alive today, then rollover the debt plus interest for 100 years, then tax everybody alive 100 years from now to pay off the accumulated debt.
Actual I = actual S is an accounting identity. Desired I = desired S is an equilibrium condition. If some stuff perishes before it is sold and consumed, we either have to include or exclude that in both I and S. Best to exclude it.
Sergei: if I don’t know who will pay the extra taxes, then I don’t know who will bear the burden. It might be workers 20 years from now, or landowners 100 years from now, or people owning bonds 42 years from now…but it’s a burden to someone, if taxes have to be increased.
David: I see those as slightly improved variants of my 1.
Got it, I guess I was just using (6) and paying it in time.
I am unsure of what you mean by “on average” and “forever” but I have no doubt 4 is true presently and for the foreseeable future.
Lord: I don’t know what I mean by “on average” and “forever” either, because my math isn’t good enough. But r being less than g for the “foeseeable future” isn’t good enough. Because if we follow that policy, the debt will still be there in the unseeable future, and it’s that unseeable future that worries me.
Nick Rowe: “Actual I = actual S is an accounting identity.”
Then why does the term, (S-I), appear in equations?
Min: because:
1. It’s talking about desired I and desired S.
2. Its’ talking about private S and I, as opposed to national S and I, which includes government S and I.
3. It’s an open economy, and S and I ignore foreign S and I.
Nick Rowe: “There is no burden on future generations because the extra taxes to service the debt will be paid to those same future generations that inherit (sic) the bonds.” That is a really bad argument.”
Who makes that argument? I have not heard it, except as a straw man. What I have observed is debt/deficit hawks claim that the national debt will be a burden on future generations. They say that in order to scare people. People get scared because they make the analogy with owing a debt to a bank or loan shark. In this analogy, the future generations are the ones who owe the debt. Who they owe the debt to is some Other. To counter that propaganda, it is correct to point out that, for the most part, the people to whom the debt is owed will be fellow citizens, just like today. Therefore the supposed burden that debt/deficit hawks are using to scare people is not the spectre that they conjure.
That is not to say that people cannot pass burdens on to future generations.
Nick Rowe: “I don’t know what I mean by “on average” and “forever” either, because my math isn’t good enough. But r being less than g for the “foeseeable future” isn’t good enough. Because if we follow that policy, the debt will still be there in the unseeable future, and it’s that unseeable future that worries me.”
There are any number of possibilities for the unseeable future. A caste society with debt peonage is one. A revolution is another. Inflation is another. If r > g becomes a problem, inflation can restore the balance between debtors and creditors. No? Boundary conditions matter.
Min: “Who makes that argument? I have not heard it, except as a straw man.”
Dean Baker: http://www.cepr.net/index.php/blogs/beat-the-press/children-and-grandchilden-do-not-pay-for-budget-deficits-they-get-interest-on-the-bonds.
Paul Krugman: http://krugman.blogs.nytimes.com/2011/12/28/debt-is-mostly-money-we-owe-to-ourselves/
“That’s not to say that high debt can’t cause problems — it certainly can. But these are problems of distribution and incentives, not the burden of debt as is commonly understood. And as Dean says, talking about leaving a burden to our children is especially nonsensical; what we are leaving behind is promises that some of our children will pay money to other children, which is a very different kettle of fish.”
I want to try what amounts to a variation on argument number 4, although it sounds quite different.
— The marginal product of capital is the rate at which we can substitute future consumption for current consumption.
— Therefore it’s appropriate to discount future flows at the marginal product of capital to get a present value.
— If you discount the flows from a government bond at (any reasonable estimate of) the marginal product of capital, you get a value that is far less than the market value.
— Therefore the government gets a windfall when it sells a bond: it is, in effect, borrowing a little bit of money (at an interest rate equal to the marginal product of capital) and having a bunch of extra money dropped in its lap (and potentially that of taxpayers, whose current taxes can be reduced, and any reduction in distortions is gravy) which it doesn’t have to pay back.
— Essentially, then, the government, merely by virtue of being a trusted sovereign, has access to a technology for producing something (liquidity? insurance?) that people want, but it can only produce this product when it borrows money.
— Therefore, any downside to deficit spending has to be set off against the fact that it enables this marvelous technology that produces something from nothing. Presumably there is some optimum, but it is almost certainly a positive deficit, even during good times, except perhaps if the debt is already very high (which presumably makes the government increasingly less trustworthy and thereby reduces the value of its product by more than the amount it makes from selling the product).
Even more than that, we can’t be sure which direction the burden is passing, to the future or to the present, so foreseeable is our best measure for our actions.
Andy: “The marginal product of capital is the rate at which we can substitute future consumption for current consumption.”
Hush! All the Cambridge UK types will come down on you like a ton of bricks!
What you meant to say is “The marginal product of capital times the marginal rate of transformation of current consumption goods into current capital goods is the rate at which we can substitute future consumption for current consumption.”
OK, now where were we?
Dunno. 4 works even in a pure consumption economy. Once we introduce liquidity premia, my mind drifts off to Friedman’s Optimum Quantity of Money. How much risk is the government taking on? Dunno. I can’t get my brain around this properly yet.
BTW, did you see my NK OLG model? Dedicated to you. Was that the argument you had in mind?
Nick,
1. I’ve posted this elsewhere, but what if we implement a bond sales tax based on the spread between the rate of interest depending on debt service as a share of GDP? How could “future generations” ever be burdened then?
2. Why should bond buyers/bond holders ever expect “burdensome” returns? How is this normal? If you want/expect high returns you should be investing in the private market.
3. Why do you think it’s right to say bond holders can be burdened since they’re taking on risk voluntarily? Nothing is forcing anyone to buy a bond.
4. A bondholder is a creditor, but the debtor is not “future cohorts,” but the economy as a whole. The economy as a whole include bondholders. Therefore bondholders cannot be burdened if they’re taxed to pay for their own bonds.
That should have been: “the spread between the rate of interest and the rate of growth,” but it would be like a reverse-inheritance tax.
a) Don’t use bad arguments to defend deficits, but equally don’t use bad arguments to avoid them. The worst argument to avoid them is that they have to be paid off – the Ricardian assumption. That’s the wrong starting point for the argument. Be proactive rather than passive in resisting the Ricardian meme. There is something to be said for the value of a process that produces debt at the risk free rate (and its term structure in the case of bonds). The government has a monopoly on risk free rate liabilities and there is demand for them (I think DeLong emphasizes this). This doesn’t mean deficits should be reckless and unlimited and it doesn’t mean surpluses won’t be justified on occasion. But the assumption of repayment is wrong as the starting point. That’s why the assumption of taxes as a conditional outcome is so central to your exposition of debt burden.
b) For today’s environment, your number 3 is good. Market pricing is basically telling you to run deficits, and to lock in long term money in doing so – which unfortunately is QE and twist incompatible.
” Therefore, any downside to deficit spending has to be set off against the fact that it enables this marvelous technology that produces something from nothing. Presumably there is some optimum, but it is almost certainly a positive deficit, even during good times, except perhaps if the debt is already very high (which presumably makes the government increasingly less trustworthy and thereby reduces the value of its product by more than the amount it makes from selling the product).”
Is that an argument for a minimum level of gross government debt regardless of the net debt of a sovereign? Providing risk-free returns as a sort of infrastructure…
So, something like Norway having a government debt and keeping assets in a sovereign wealth fund, even though its net debt is negative.
anon:
1. Hmmmm. I’m not sure I fully understand this. Wouldn’t there be a risk the government would be suddenly unable to borrow if it hit that ceiling, because it would have to raise rates to compensate for the tax? Or is it like my trill perpetuities plan?
2 and 3. It’s not the bondholders who pay the burden. It’s the people who pay the taxes to pay the bondholders.
4. See my counterexamples.
JKH: we don’t know the future, and don’t know whether, at the margin, the present value of extra taxes will equal the extra deficit.
I understand that economic models with inheritance typically satisfy Ricardian Equivalence, but that does not mean that all models with inheritance satisfy Ricardian Equivalence. Govt. bonds could be inherited with Ricardian Equivalence being violated for other reasons (for instance, the poor, who do not own govt. bonds are liquidity constrained). A realistic model (well, to the extent macro models can be realistic) could be easily constructed with dynasties of wealthy bond owners and overlapping generations of non-Ricardian poor.
While I understand your reasoning, every time you conflate the concept of inheritance and Ricardian Equivalence, I feel you are making a false equivalence. This bothers me even more in posts where you are being very careful and specific about various conditions.
Maybe rates would rise. But we’re talking about burdening the future, if a bond auction fails and the government can’t borrow, it can’t burden. But you’re not answering why bond holders should be expecting to extract burdens on the economy in the first place? (I’ll have to read the trill post again later, I read it before and some of it went over my head.)
2., 3., and 4. I’m saying we tax the bondholders to pay the bondholders. And that this is a risk of buying bonds. (I don’t know what you want me to see in your counterexamples, I understand the basic argument, and agree that the burden can be imposed, but am saying it’s a policy choice/risk of buying bonds, not a function of the debt.)
You can’t transfer the burden to the future by selling your bond because the person buying the bond is assuming that risk by buying the bond of their own free will. If for some generation government policy favors bond creditors that’s a policy choice. Saying that bond holders deserve to impose burdens is creditor bias.
primed: I see your point, but I’m not sure how easy it would be to construct a model where RE is false but there is no burden.
“A realistic model (well, to the extent macro models can be realistic) could be easily constructed with dynasties of wealthy bond owners and overlapping generations of non-Ricardian poor.”
A simple 50-50 model (half Ricardian population, and half making no bequests) wouldn’t do it. The deficit would need to be twice as big to cure the same recession. So future cohorts of the non-Ricardians’ kids would be twice as worse off.
anon: “I’m saying we tax the bondholders to pay the bondholders.”
OK. So I borrow money from you. I then say I’m going to tax you to pay you the money I owe you. You are worse off. (BTW, that is exactly what is happening in my model already).
Nick,
I agree that’s what is happening in your model, but you’re ignoring my argument. (I’m interpreting what we both agree is happening differently.)
“OK. So I borrow money from you. I then say I’m going to tax you to pay you the money I owe you. You are worse off.”
Notice that “you” in this case is the government. Not me, anon. When you lend the government money, the economy as a whole becomes your debtor. The economy as a whole includes you. If you’re taxed to pay some of your own bond payments, that’s simply a risk of buying a bond in the first place.
How about if all taxation is from the wealthy (those who inherit and bequeath the bonds) and deficits arise from government spending on public projects?
Sorry, “I” is the government in your example.
@ Nick Rowe
I guess you are disagreeing with me, but I do not see the evidence. Take part of the Krugman quote:
“not the burden of debt as is commonly understood.”
You are not talking about the burden of debt as it is commonly understood. Krugman is. He is not saying what you think he is.
How did you proceed from showing that debt can be a burden on future generations, to concluding that debt is a burden on future generations? The latter is an empirical issue which has not been addressed by the recent debates.
This can’t be how macroeconomics work, is it? Where is the empirical evidence?
Dan,
Nick has always said there are conditions under which debt won’t be a burden.
Could you elaborate more on point 0 or post some links where one could look deeper into the argument against inheritence?
I’m a #2, #3 guy myself.
Nick,
You would be my personal hero if you can either (a) explain or (b) get Dean Baker to clarify what he meant when he posted this in the comments of his last post on this topic:
It’s the fraction of the spending from interest on government bonds that is spent by people without kids (i.e. not saved) and not received from taxes paid by people who also owe bonds. [The true measure is the consumption based on bond wealth, but I’m going to be sloppy and use this as proxy.] Currently, net interest is a bit over 1 percent of GDP. It also only applies when the economy is at full employment (don’t hold your breath on that one). My guess is we’re talking about a burden that is hovering around 0.1-0.2 percent of GDP…
I am on the cusp of getting what he means–it’s on the tip of my tongue, as it were. But I’m not quite certain. My suspicion is that he’s wrong (i.e. not setting up the measurement correctly), and that I could do one of those big apple models to show that his measure would show little burden, when in fact the burden is huge.
But, before wasting my time on that, I want to be crystal clear on what Dean Baker means by the above. Any thoughts?
Bob,
He’s saying you have to subtract the taxes bondholders had to pay themselves to pay interest from the amount they spent and did not give to their kids. He’s “guessing” this is about ten to twenty percent of net interest payments. I’m guessing net interest is interest minus growth.
(I’m guessing the bondholders taxed would, in principal, have to be same as those who spent.)
Nick,
I will try a different tack. Suppose that there are no government bonds traded in a secondary market. However, there are government “deposit accounts”, that anyone can open by depositing funds in (the sole) government bank. This is the only bank. The savings accounts pay an overnight interest rate as set by the central bank.
In this model, there is no difference between bonds and deposits. Government spends money by marking up someone’s deposit account (whoever it buys goods from) and then immediately paying interest on it. Government taxes by reducing someone’s deposit account.
In this model, does expansionary monetary policy (e.g. fiscal policy) cause a burden on future generations?
Bob: All he is really saying is: if Ricardian Equivalence is half true, then the burden of the debt is only half the debt. Because half the bonds weren’t bought but were received as gifts from the parents. Trouble is, if Ricardian Equivalence is half true, you need double the deficit to get the same fiscal stimulus. So you end up with exactly the same burden on the kids as you would if Ricardian Equivalence were 100% false.
It’s the same old story. He wants to have it both ways. He wants to lean pro-Ricardian when it comes to the burden of the debt, but anti-Ricardian when it comes to the power of deficits in increasing AD.
Dan: It depends. Suppose we were talking about case 2, where we borrow to build a school. IIRC (I probably don’t) the empirical estimates are that it costs about $1.30 in burden ($1 direct, and $0.30 in excess burden due to distortions and collection costs) to raise $1 in tax revenue at the margin. That’s on the cost side. On the benefit side, some very brave economists do try to estimate empirically the benefits of schooling. The simplest method is to estimate the extra earnings of those with extra education, by running a regression of earnings on years of education. But we know that will overestimate the benefits, because it might be that the brighter kids, who will earn more anyway, stay longer at school, both to signal that they are bright, and because they are bright enough to be allowed to stay. And some very hard-working empirical economists try all sorts of clever tricks (using natural experiments etc.) to try to get around this problem.
So, if I thought I would be any good at doing this, I would research all the empirical public finance literature to check my memory of whether that $1.30 estimate of the costs is correct, then plough through all the empirical education literature to get an estimate of the benefits, subtract one from the other, cross my fingers, and give you an empirical estimate of the net benefits/burden.
But that’s not my comparative advantage.
rsj: “In this model, does expansionary monetary policy (e.g. fiscal policy) cause a burden on future generations?”
Yes. Because your model is formally identical to my model. I call them “bonds”; you call them “deposits at the government savings bank”.
I’m sorry for repeating myself but I’m really curious about the following possibility where ‘Ricardian equivalence’ is half true in a way that differs from your suggestion where half of the otherwise homogeneous population is Ricardian and the other half is not.
How about if all taxation is from the wealthy (those who inherit and bequeath the bonds) and deficits arise from government spending on public projects?
primed: If half the population is Ricardian (leaves bequests) and the other half is non-Ricardian (leaves no bequests) it is not obvious (to me, right now) that the economy as a whole will be half-Ricardian (even if those two halves are otherwise identical). That’s because the Ricardian half may end up saving more than their transfer payment if the non-Ricardian half doesn’t have any kids, so the Ricardian’s kids will bear the whole burden, not half the burden.
When I was younger I think I got my head around this once, after reading a good paper by Willem Buiter. Now I’ve forgotten what I once figured out, and can’t figure it out again from scratch. (This will happen to you too, one day.) But it may come back to me.
That’s why I wimped out in my response to Bob Murphy, and just assumed the economy is half-Ricardian, so that half the bonds are bequeathed, without saying what that meant in terms of people and kids.
Nick: “So, if I thought I would be any good at doing this, I would research all the empirical public finance literature to check my memory of whether that $1.30 estimate of the costs is correct, then plough through all the empirical education literature to get an estimate of the benefits, subtract one from the other, cross my fingers, and give you an empirical estimate of the net benefits/burden.”
I’d say the 1.30 # is close enough for jazz. The “marginal cost of funds” (the technical name for the 1.30 #) varies a lot depending upon how the taxes are raised – the marginal cost of consumption taxes may be less than the marginal cost of taxes on investment income, for example, or vice versa. It wouldn’t take long to survey the literature – Bev Dahlby has a recent book that looks at it all.
On the benefits of education, though, I think it’s really hard to get solid numbers, for the reasons that you mention in your comment. Yes, we can observe how much more, historically, the educated have earned than the less educated. How much of that difference is causally linked to education is really hard to find out (some people have tried to use changes in compulsory schooling legislation to try to find out, but that just gets at K-12 education.) What the returns to education will be going forward is another question entirely, and it will be years before we’ll know.
Ricardian equivalence is sometimes true and sometimes not. This is well demonstrated empirically.
So to make an objection on grounds of principle, you need to lay out why you think the empirical evidence is wrong.
Dan: you see, there are good economists, like Frances, who both do empirical work and who have knowledge of the empirical literature. If all economists were like me, economics would be rubbish. Fortunately they aren’t all like me. And it’s OK for me to specialise in what I do best, because I know there are other economists who specialise in doing all the good things that I don’t do.
Worth noting as well that all these arguments about debt being a burden on future generations apply equally well to digging oil and coal and other consumables out of the ground and using them.
Metatone: Damn! Yes, I forgot to mention Ricardian Equivalence in this post. That’s because I have spent the last 10 posts on this subject arguing with people who are convinced RE is wrong, but there is no debt burden, because “future generations inherit (sic) the bonds too”!!
My sense of the empirical literature is that RE is roughly half true. That sounds roughly plausible to me, though I would say that the extent to which RE is true depends on a lot of things.
Metatone: yep on the oil and coal. One big benefit though, that often gets forgotten, is that we also bequeath technology to future generations. And once the patents have expired, that is a true bequest.
@Nick: re #2. Speaking as a non-economist, this makes sense if what gets built is roads, sewage and water improvements, bridges, etc. But since you are looking at bad arguments, we all know that (1) the total costs of operating a school, hospital, etc are many times their construction costs (staff and equipment, to say nothing of value for money and output); (yes I know bridges have operating costs too); and (2) one doesn’t have to look far for examples of “infrastructure” funds that ended up being used for such vital economic drivers as statues of dead politicians, tap-dancing museums, and subsidies for yak ranches in New Brunswick.