(I had been planning to write this post before Steven Landsburg started a whole blogosphere argument about what taxes are for. Honest!)
Functional Finance says you only use taxes if you want to reduce Aggregate Demand to prevent inflation. The Long Run Government Budget Constraint says you use taxes to pay for past, present or future government spending. They sound very different. They aren't.
There's a general principle in economics: first you eat the free lunches; then you look at the hard trade-offs. Functional Finance says "first eat the free lunches". The Long Run Government Budget Constraint says "then look at the hard trade-offs".
Suppose, just suppose, that if you kept on doing what you were planning to do, you never had to worry about inflation. Not now, not in the future, not ever. Because Aggregate Demand was too low now, and was projected to be too low forever. So you are not worried about inflation. Instead you are worried about deflation. And you were a government that could print your own money. What would you do?
You would print money and spend it. Or print money and use it to finance tax cuts. And you would keep on doing it, more and more, until you got to the point where you did start to worry about inflation. You first eat all the free lunches.
That's the underlying kernel of truth in Abba Lerner's Functional Finance (pdf) [see also his book The Economics of Control (pdf)]. And I don't know of any mainstream macroeconomist who would disagree. You use taxes only so you don't have to print money. When you get to the point that Aggregate Demand is high enough, so you start to worry about inflation, you use taxes to finance past, present, or future government expenditure precisely because you don't want to print more money and make inflation higher.
Suppose inflation isn't a problem right now, because Aggregate Demand is currently too low. Does that mean the government should print money and spend it? Not necessarily. Print money yes, but instead of spending it on goods, or on tax cuts, it might be better to use it to buy back some interest-paying government bonds. Because even though inflation isn't a problem right now, it may be a problem some time in the future. So you can buy the money back in future, by re-issuing the bonds (and save on interest in the meantime) without having to raise future taxes or cut future spending.
Here's an easier way to think about it. Print enough money to get Aggregate Demand and inflation where you want it to be. That's the free lunch the government can eat. Any additional government spending must be paid for, sooner or later, with taxes. The present value of taxes, plus the present value of newly-printed money (seigniorage), equals the present value of government spending, plus the existing debt.
Seigniorage revenue belongs in the government budget constraint. The government can pay for part of its spending by printing money. But don't get too excited. It's not that big, on average. If central bank currency is around 5% of annual nominal income, and if nominal income is growing at 5% per year (3% real plus 2% inflation) then 5% x 5% = 0.25% of GDP. (In the right ballpark for Canada — it shows up as the profits the Bank of Canada hands over to the government — but maybe double it for the US). Printing money is a nice little sideline, but we are still going to need taxes.
Let's ask a slightly different question. Why do governments pay interest on their debt? Actually, it sounds like a different question, but it's really the same question. Why finance government deficits with interest-paying debt, when you could use non-interest-paying currency?
The answer is the same: you pay interest on the debt to encourage people to hold it and stop people spending it. If you cut the interest rate on government debt, will people sell it back to the government for money, spend the money, and cause inflation? If not, then Aggregate Demand is too low, and the problem is deflation, not inflation. So there's a free lunch from cutting the interest rate on government debt. And the government should eat that free lunch. And then the Long Run Government Budget Constraint kicks in again.
Now suppose the real rate of interest on government debt is below the real growth rate of the economy. (Or the nominal rate of interest is below the growth rate of nominal GDP — same thing). And suppose it will be like that forever, if you keep on doing what you were planning to do. The government can run a Ponzi scheme forever. It can borrow and spend, then borrow to pay the interest forever, and the debt grows more slowly than the economy, and the debt/GDP ratio declines over time. The Long Run Government Budget Constraint is undefined. The Present Value of taxes can be less than the Present Value of Government spending.
That's another free lunch that needs eating. The economy is dynamically inefficient. The economy wants a Ponzi scheme. And the government should satisfy that demand. Issue debt until the interest rate equals the growth rate. Then the Long Run Government Budget Constraint kicks in again.
Worthwhile Canadian Initiative 
Vimothy,
The Model SIM is too simple and is a way to build complicated models .. one by one …
G&L models are frameworks and one is bound to find something debatable. More importantly the parameters which have been assumed to be fixed to start with such as propensity to consume bring more life one one starts thinking of them being changing all the time – a point mentioned in the book as well.
As far as wealth is concerned, they do not take future income into account in the definition of wealth, except for financial securities which have a market value because interest income is paid. But that is the spirit of national accountants as well. True, the latter are just measuring stocks and flows and not the future… but the important point in those G&L behavioral assumptions is that sectors on the whole do not care .. these have more to do with animal spirits which can’t be measured and so forth.
There is a dependence of expectation of incomes and if you see some consumption function have expected income instead of the income. In the simplest case it is assumed to be the previous period income but one is free to choose something else such as 5% more than the previous period income. The models are written truly in the Keynesian spirit of uncertainties etc. You can give the propensity to consume a time dependence. The goodness of the models lie in the openness and the attitude of the authors is completely different from the ones you meet online 🙂
The models are one end of abstraction but in fact if you read Wynne Godley’s articles at Levy, he is seen trying to write about prospects for the US economy which is more about policy. Which brings me to the sectoral balances… you can learn more about them by reading his articles patiently. For example nowhere does he assume that it was governments forcing people to go into debt and that the severe day of reckoning the US was heading into was due to the irresponsible behaviour of all sectors of the economy including the foreign sector.
To me the idea of aiming a surplus is a slightly shady argument. The Clinton government went into a surplus not because of its own choosing but because of capital gains due to the dot com boom. The argument presented by Godley is that given the nature of such an attitude and with the CBO projecting a retirement of debt over the next 15 years (around 2000), the government’s fiscal stance is going to be tight and if the propensity to save increases, it will cause a recession given that it may come as a surprise.
However his attitude was also that the economists assign a very limited role to the government and this is not the right path.
Also, one thing missed is what happened after the short recession around 2001 … so you can read in Wynne Godley’s articles at Levy that the US government actually relaxed fiscal policy by a humongous amount .. and that itself led to another process which led to further sectoral imbalances. So there is a story about the huge relaxation of fiscal policy which itself added to the cracks in the foundation of growth by contributing to imbalances in the private sector flows and the balance of payments.. the story which you won’t find in the blogosphere.
I understand the points in your comment and you have to be careful in distinguishing arguments which seem superficially the same.
On, the other hand, there is an overkill in MMTosphere and this overkill is counterproductive.
Beowulf,
Great you noticed that.
True that was the Bretton Woods era and there is a balance of payments constraint. However to understand it itself needs a good description … it doesn’t matter NAFA=PSBR+BP and the dynamics is quite similar. Gold loss can be cured by borrowing from abroad if allowed by foreigners.
The present era is a Bretton Woods era without the gold and flexible exchange rates. The balance of payments constraint remains and endemic deficits can remain as long as foreigners allow just like any other institutional setups.
Abba Lerner was a friend of Nicholas Kaldor and the full description of the balance of payments constraint comes from Nicholas Kaldor – explained in detail by his followers.
Nicholas Kaldor and Lerner would have had influences on each other surely. This is what Kaldor wrote in 1941 before functional finance was being written:
It is impossible to judge intelligently the system of taxation, or the scale of public expenditures, without a quantitative record of the total economic activity of the nation, which forms the background. This is perhaps even more important in war-time, when the Government controls so much larger a part of the national income; but it is vital in peace-time as well. If a statement of this kind had been presented year by year, simultaneously with the Budget, many financial mistakes of
past Governments might have been avoided. Moreover, the regular publication of this document would
stimulate both Government and Parliament to look upon the level and the stability of the National Income, rather than the conventional and narrowly financial standards, as the true criterion of
budgetary policy; to regard the movements of the national expenditure, and not merely of the expenditures of public departments, as within their province. It is on the assumption of this wider responsibility that our best hope lies for the post-war world.
-‘The White Paper on national income and expenditure’
I vow to find something written by Abba Lerner on the balance of payments superconstraint in the Post Bretton-Woods era!
Min: “Don’t some mainstream economists believe that it is, that the long run limit of G-T = 0? Not only that, don’t they believe that that belief is rational?”
Hard to do math on Typepad (hard for me to do math anywhere), but:
Start with the short-run GBC (ignoring seigniorage, just for simplicity):
G +rB = T + B next period – B
Re-write that same equation for next period, solve for Bnextperiod, substitute it back into the first, then repeat for t periods, and you get:
B + PV(G up to t) = PV(T up to t) + PV(B in period t)
Now, we need to take the limit of that equation as t goes to infinity. If the growth rate of the economy is less than r, so that G, T, and B can’t keep on growing faster than r forever, the answer is straightforward:
B + PV(G) = PV(T) (the weird term at the end must go to zero in the limit). That’s the long run government budget constraint. Add in the PV of seigniorage if you wish.
But if the growth rate of the economy exceeds r, it’s possible for the growth rates of G, T, and B to exceed r, so the Present Values are all infinite. That’s when things get weird. Essentially, the economy is in a Ponzi zone.
I don’t think Lerner’s ideas on foreign trade are essentially different from “MMT”, or is in a Ramananigm. It has some terminological difference – I had the Ramananigm impression at first. But imho Lerner is often clearer- of course foreign savings of dollars, in whatever form, constitute a foreign debt of the USA, as Ramanan rightly insists. But just as the modern MMTers, Lerner suggests using functional finance to offset demand leakages, foreign or domestic, and holds they are not essentially different. The Ramananigm is roughly what Lerner calls “sentimental internationalism” in his Economics of Employment, which he wrote, perhaps following Keynes’, for “those at the gate” rather than “the cognoscenti in the temple”. Both books are awesome.
Ramanan, as you know, Lerner developing the Lerner Symmetry Theorem in the 1930s, but don’t know if he did any work on post-Bretton Woods trade.
http://en.wikipedia.org/wiki/Lerner_symmetry_theorem
To save a linkthrough on the “in-Ramananigm” Trump comment (apologies to Nick for going OT a moment). The Donald has made surprisingly MMT-friendly statements on fiscal policy with one notable exception that I figured would warm your heart… “The centerpiece of his deficit reduction program appears to be a 25 percent tax on all Chinese imports… Trump also sees no need to reform entitlements, which will magically attain solvency on their own. “When this country becomes profitable again, we can take care of our sick; we can take care of our needy,” he told Human Events. “We don’t have to cut Social Security; we don’t have to cut Medicare and Medicaid. We can take care of people that need to be taken care of. And I’ll be able to do that.”
And he says we won’t need to raise taxes either. Trump is suggesting that, as our economy improves, it will expand to cover trillions of dollars in future deficits… (link above)
Instead of a “China tariff” it’d be better if he were pushing, how did Wynne Godley phrase it, “nonselective protectionism matched with fiscal relaxation” (i.e. cutting payroll tax) ,but I guess we’re all works in progress. To make the initiative worthwhile to Canadians, of course, we should exempt our NAFTA partners (and back on topic!). :o)
Nick @ 6:47,
re: ” It makes no sense for households to pay a premium for bonds over capital if they do not consider bonds to be wealth.”, by “households” I was referring to the household sector.
Obviously an individual household in either world would value the government bond at some non-zero amount, as it delivers a stream of income. However, in the Ricardian world, the household will not pay a premium for the bond, whereas in the non-ricardian world, the household will pay a premium for the bond.
In terms of wealth, the future tax obligation is contingent on nominal income variations, and the return from capital is as well, but the income stream delivered by bonds is not, so households would prefer to have the variable tax obligation and the bond, rather than not having the bond or the tax obligation. Because they prefer it, they pay a premium for it, and because they pay a premium, the future tax obligation is reduced as well.
The economy need not be dynamically inefficient, or in a ponzi mode as you put it, as the return demanded of capital is higher than the return demanded of risk-free bonds, and the private sector, or at least the non-financial private sector, cannot borrow at the risk-free rate. They can only lend at the risk free rate.
This was Abel & Summer’s argument. The capital stock is earning a higher rate, which can be greater than the growth rate of the economy. Low risk-free rates only tell you that households are credit constrained and risk-averse, they don’t tell you anything about dynamic efficiency per se.
Calgacus,
Would verify that from original articles rather than relying on someone’s account.
Nick Rowe: “Hard to do math on Typepad”
Thanks for the math, Nick. 🙂
RSJ: sorry. I had totally misread you.
Hi Nick,
“But if the growth rate of the economy exceeds r, it’s possible for the growth rates of G, T, and B to exceed r, so the Present Values are all infinite. That’s when things get weird. Essentially, the economy is in a Ponzi zone.”
Now is the time to go to Moslers website and really understand the monetary operations of the fed and treasury. Now.
First, understand how the fed and treasury work, why debt = money. Then read Warren’s “the natural rate of interest is zero.” We choose the interest rate we pay on the debt. If you doubt me, see Perry Merhlings post about how the fed pegged interest rates throughout the WWII. http://ineteconomics.org/blog/money-view/the-new-federal-reserve#comments
You are so close! Welcome. When you get here, you’ll find all your old skills will actually increase in value, and there is much work to do. And remember to pass the map to Brad Delong!
Rmanan:
Oh and I meant to say:
Steve,
I think wages rose a lot during the 70s and hence prices. And hence wages and hence prices.
If wage bargaining gets troublesome, and workers demand more wages, firms need to price higher to recover the costs. This may result in a higher wage bargaining due to rising prices … and dynamics along those lines.
The reason money supply is associated with prices is that in that period, I believe (no data with me) that the two rose together rapidly. So the Monetarists thought that the price rise was caused due to rising money supply (and only partly accepting it has something to do with labour unions). The reason the money supply rose was that more loans were taken by firms to pay higher wages … (the reverse causality).
I haven’t done any proper analysis on this, but there seems to have been an unfortunate indexation going around at that time as well.
What r < g means is that given any primary deficit/GDP bound, there is a (finite) Debt/GDP bound that is a function of the deficit to GDP bound.
Right, exactly. This is a surprisingly elementary mistake on Nick’s part.
You can follow the procedure I described above, or the basically equivalent procedure of the original post, without any information about the current debt/GDP ratio. And you can follow it forever, and the debt/GDP ratio will converge on the same finite value, not matter where it is presently.
Reread your original post, Nick. Where does it say “Now look at the current level of public debt and change your behavior accordingly”? That’s right, it doesn’t. We set today’s government deficit based on today’s macroeconomic conditions. The debt-GDP ratio doesn’t matter. That’s the key takeaway from functional finance, and the fundamental difference from the mainstream view. And you’ve actually shown why functional finance is right and the mainstream is wrong, you just don’t realize it yet.
Sorry, the statement quoted is correct, but it isn’t quite what I met. If you are not concerned with the primary deficit but with the total deficit, then the debt/GDP ratio stabilizes at d/g (where d is the deficit as a share of GDP and g is the growth rate of GDP), regardless of the initial debt level and regardless of r. I think part of the confusion here arises because people are used to talking about the primary deficit, but functional-finance arguments are in terms of the total deficit.
It may be that following the rules of functional finance causes interest payments to rise as a share of GDP. But so what? Either inflation remains below target, in which case you haven’t hit the long-run constraint and should increase borrowing more; or else inflation rises above target, in which case the rules of functional finance tell you to raise taxes regardless of the level of interest payments or debt. In other words, interest payments are just another addition to income for the private economy; and just like any other increment to private income, if they result in AD running ahead of potential output, then you raise taxes.
The larger point — which unfortunately it’s probably too late to expect a response on — is that the argument of the original post directly contradicts Nick’s later claim that an exogenous increase in debt, all else equal, implies a lower level of spending or higher level of taxes.
On the Lerner comment:
In the ’70’s I had a professor who was a colleague of both Allais and Debreu.
For some reasons we asked him what he thought of Joan Robinson eventually getting the Prize. He reacted in horror: ” We won’t let that happen!”. She may have not deserved it but she was not ” one of us”.
JW: “The larger point — which unfortunately it’s probably too late to expect a response on — is that the argument of the original post directly contradicts Nick’s later claim that an exogenous increase in debt, all else equal, implies a lower level of spending or higher level of taxes.”
Not too late. I’m still following. But too late for a response today. It’s been a busy tiring day for me, and so my brain is not in the best shape to try to explain clearly.
Ramamman wrote:
I think wages rose a lot during the 70s and hence prices. And hence wages and hence prices.
If wage bargaining gets troublesome, and workers demand more wages, firms need to price higher to recover the costs. This may result in a higher wage bargaining due to rising prices … and dynamics along those lines.
The reason money supply is associated with prices is that in that period, I believe (no data with me) that the two rose together rapidly. So the Monetarists thought that the price rise was caused due to rising money supply (and only partly accepting it has something to do with labour unions). The reason the money supply rose was that more loans were taken by firms to pay higher wages … (the reverse causality).
I haven’t done any proper analysis on this, but there seems to have been an unfortunate indexation going around at that time as well.
I have a different view. There is more than one kind of recession. I will say that again, there is more than one kind of recession. I have a little article demonstrating a Hayekian graphical model of the economy which I got from Roger W. Garrison in his book “Time and Money”.
Before anybody screams at me for reading it, you can make Garrison’s model behave in a completely Keynsian way by altering a few of the graphical interpretations and just applying Keynes’s assumptions wholesale. I accept Keynesian economics but my little model shows how it can come up short. I like my model because it explicitly includes time and money in the macroeconomy and breaks out the supply side in more detail than is typical.
The heart of the model is the Production Possibilities Frontier, a curve on a graph of Consumption vs. Investment. The Investment axis is reflected down to break out the Funds Market, a graph of interest rates vs Investment. When an economy is on the PPF you get a stable, “natural” rate of interest with the money supply. You can draw two boundaries on the Funds Market to determine the total amount of money available for investment and consumption.
Anyway, recessions in my model can happen in two ways:
1: A net contraction of the Production Possibilities Frontier due to a supply shock. This will drag consumption and investment down and raise unemployment. This is what I call a Hayekian recession. It is caused a a shorted of real goods. This is what happened in the 1970’s. Several books of the period detail that first wheat shortages developed after crops failed in China and Russia and the world had a grain shortage.
The United States had an oil supply problem in the 1970’s as well. Until 1968 the US produced enough oil domestically to satisfy domestic consumption. By the 1970’s these supplies were exhausted and the US turned to Persian Gulf and Venezuelan oil. For real and political reasons, the US had to expend more resources and more money to purchase this oil.
These shocks contracted the US Production Possibilities Frontier. The result was unemployment as industry contracted and also price increases as the economy adjusted to oil shortages. A steady money supply with a decreasing PPF implies price increases (same amount of money chasing less goods).
This is how you get unemployment and price increases occurring together, stagflation. This is the story of the 1970’s.
This isn’t the only kind of recession. Which brings me to
2) A Keynesian Recession.
Instead of the Production Possibilities Frontier contracting in response to a supply shock, assume the PPF stays steady. But contract the money supply. According to Keynes consumption is more steady or sticky than investment, so investment decreases. Remember that natural rate rate of interest that the [C,I] point on the PPF maps so nicely to on the funds graph? Well, that point is now BEYOND the Money Supply Frontier. Just as Keynes said, no amount of interest rate adjustment will get to recovery in this situation.
If the money supply contracts, it will drag the economy to a new [C,I] point off the PPF. A gap will open up, a production gap. If consumption starts to fall it will devastate the economy (modelled as a Hayekian triangle with stages of production), giving every stage a haircut. In real life it looks like random business failing for lack of demand, money problems or other non-supply related reasons.
Because the money supply contracted, we don’t have enough money to accommodate all monetary transactions in the economy anymore. We get a demand shortage as uncompleted but desired transactions pile up. This is the story of the 1930’s and the current Financial Crisis.
At root, and this is my own interpretation, my model has confidence-based money. Money isn’t just notes, it’s a product of the banking system and of companies reinvesting funds. The money supply is based on confidence because it is a product of financial actors (banks) who engage in loan activity where confidence in repayment is everything. If you get a panic it will cause widespread devastation. Sound familiar?
Roger Garrison would probably shoot me for giving his model the second interpretation, but that’s life. So we now have two consistent and different modes of economic failure. One we went through in the 1930’s and are in today, and one we experienced in the 1970’s. Different problems, different causes, different solutions, same economy.
Nick: “But, if the banking system is competitive, the super-normal seigniorage profits of the commercial banks get competed down to zero, either eaten up in operating costs, or by paying interest on demand deposits.”
I’d like to follow up on this debate between Nick, Vimothy, Andy and RSJ. This is one of the issues where I’m in almost full agreement with RSJ. Here is the main reason I disagree with you, Nick: the marginal cost of lending for a bank is independent of the rate on demand deposits: Term loans are benchmarked off term wholesale borrowing. Overnight lending (lines of credit/revolvers) are benchmarked off the interbank lending rate. The reason is that the liquidity of the funding source must match the liquidity provided to the borrower. If a customer draws a billion dollar revolving LOC, the newly created deposits flow to the banking system in general; not to the lending bank. Therefore the revolver must be financed off the interbank market that the lending bank can draw on on demand. There is no mechanism whereby the zero rate on demand deposits ever makes it’s way into the pricing of a loan.
And Canadian demand deposits and most overnight savings deposits do not pay interest (or maybe 5 basis points on ON savings). I believe the reasons in the case of demand deposits are regulatory, but regardless, the supply of demand deposits is highly inelastic as a function of the rate paid on those deposits. The aggregates are huge but holdings only exist for a very short period in any given account, so people ignore the rate. They simply keep their checking accounts with the institution that provides their lending/brokerage/investment banking etc services. So demand deposits are a bit like land: a commodity whose supply is fixed (or at least independent of price) and which is a positive externality to the beneficiary, in this case the banking sector.
So RSJ is right: it should be taxed (as should land value) at the risk free rate. This would be a highly efficient tax with little or no impact on the cost of credit. Of course, they should also pay the fair market price of deposit insurance. A better alternative would be for the BOC to provide interest bearing transactional accounts to anyone, effectively just extending the service (LVTS) that they currently provide only to the commercial banks. Then revoke deposit insurance and let the banks raise capital at free market rates.
And as for Andy’s point that the cost of transactions is high, it’s simply not true. Back in the days of stage coaches and banditos, surely. But today it’s a joke. Just look at the cost of executing stock trades for example (small fractions of a cent). And none of the costs are proportional to the size of a deposit. Look over any banks income statement. Try to find the expense item for payment systems. It’s negligible and entirely funded by transaction fees.
K,
That’s an excellent point, and one which I should have made earlier.
It is just like fixed and marginal costs, but each bank has a fixed “benefit” corresponding to interest income from supplying deposits.
This benefit is not variable with the amount of loans made — each individual bank does not increase its (level) of deposit liabilities by granting a loan.
Therefore banks will continue to make loans up until the marginal costs (e.g. FedFunds + capital requirements, etc.) are the marginal benefits. They will not then charge an additional discount or supply more loans to remove the fixed benefit.
Price competition cannot reduce the fixed benefit, which is the aggregate level of deposits. Taxes should remove that benefit.
Trader’scrucible: “Then read Warren’s “the natural rate of interest is zero.””
I did read it, and commented on it somewhere in comments to my last post, on MMT. Let me re-cap what I said:
If you believe the IS curve is vertical, because desired saving and investment are independent of the rate of interest, then: lowering the rate of interest will not increase AD; and the natural rate of interest is undefined. The rate of interest no longer serves any intertemporal allocative function in such a model; it merely re-distributes income between debtors and creditors. Fiscal policy has to bear the whole responsibility of AD management; and the central bank might as well set the rate of interest to zero.
That, to my mind, is the underlying assumption of MMT in general, and for Warren’s argument for setting r=0. (I think Warren’s way of expressing this argument by saying the natural rate is zero just confuses things, because he is not using the words “natural rate” in the same way the rest of us are).
It’s reminiscent of Milton Friedman’s theoretical argument “The Optimum Quantity of Money” for setting a nominal rate of interest at zero. Except that Friedman saw the natural real rate as positive, and wanted to get the nominal rate to zero through steady deflation.
But if you take the standard assumption that the IS curve slopes down, Warren’s argument doesn’t work. Because on average setting r=0 would mean very high AD, and so you also have to use a very contractionary fiscal policy to shift the IS curve left, making the natural rate zero, to keep AD from being too high. So the government debt would go to zero, and then get more and more negative.
JW: “In other words, interest payments are just another addition to income for the private economy; and just like any other increment to private income, if they result in AD running ahead of potential output, then you raise taxes.”
That is the key assumption of MMT — that interest rates merely re-distribute income between debtors and creditors. But that ignores the substitution effect of interest rates on desired investment and consumption — on intertemporal consumption and production decisions — that standard theory emphasises. If r is set to keep AD on target, and ir that requires setting r at or above g, then the Long Run Government Budget Constraint follows from standard accounting plus a bit of algebra, plus the assumption that the debt/GDP ratio must be finite:
Existing debt = PV(taxes)+PV(seigniorage)-PV(government spending)
“The larger point — which unfortunately it’s probably too late to expect a response on — is that the argument of the original post directly contradicts Nick’s later claim that an exogenous increase in debt, all else equal, implies a lower level of spending or higher level of taxes.”
And it follows immediately from that accounting identity that if you suddenly find the existing debt is higher than you thought it was, the left hand side is bigger, so the right hand side must be bigger too. You must either increase taxes, or seigniorage, or cut government spending, either now or some time in the future.
The tricky bit is working out what happens when r is less than g, because then the algebra used to derive the LRGBC above doesn’t work.
Because on average setting r=0 would mean very high AD, and so you also have to use a very contractionary fiscal policy to shift the IS curve left, making the natural rate zero, to keep AD from being too high. So the government debt would go to zero, and then get more and more negative.
I think one must always read the MMT r=0 arguments together with the corresponding policy proposals for the financial sector. One can’t happen without the other. 90% of the control/stabilization would be via shrinking of the sector and not via ‘interest rate discipline’ – which MMT sees as blunt tool because it is economically ambiguous and socially questionable. The whole r=0 edifice rests upon a severely stunted banking sector that functions as private-public enterprises with a clear charter to pursue ‘public purpose’.
Assuming for a moment, as economists like to do, that MMT=Mosler, here are his proposals for the US:
Click to access toronto.pdf
OK, so what about the case where r is less than g? (BTW, try to avoid the less than or greater than symbols, because one of them causes Typepad to freak out for some reason).
The standard model most economists use to explore this case is Samuelson’s 1958 overlapping generations model. I discuss that world in this old post:
http://worthwhile.typepad.com/worthwhile_canadian_initi/2010/03/do-we-need-a-bubble.html
The basic idea is that the economy is dynamically inefficient. Even at full employment. It would be possible to make the current old generation better off by transferring resources from the next young generation, and so on ad infinitum, without making any generation worse off. The economy needs a bubble, chain letter, Ponzi scheme (which are all the same fundamentally). Government debt could be that Ponzi scheme.
But, this does not mean government revenue from the Ponzi scheme is unlimited. As the debt/GDP rises, the demand for the savings vehicle is satisfied, and r rises until it equals g. There is only so much debt the young generation is willing to buy from the old generation, and that depends on r, so as the debt/GDP ratio rises, r rises until it is equal to g. (Clearly, if the young simply cannot afford to buy all the debt from the old, even with all their income, we have gone far past that limit.)
The revenue the government can earn from a Ponzi scheme is bounded, even if r less than g so the Ponzi scheme is stable, because as the size of the Ponzi gets larger, r rises until it equals g, and the Ponzi scheme becomes unstable.
Here’s a follow-up post on the demand and supply of bubbles, in case anyone’s interested in seeing more:
http://worthwhile.typepad.com/worthwhile_canadian_initi/2010/03/the-supply-and-demand-for-bubbles.html
BTW. A couple of days ago an MMT blogger said I was “60% there” (or something like that) with this post. I was too busy to respond at the time, but now I can’t find it. Anyone know who it was? Thanks.
K: Presumably, the administrative costs to the commercial bank of providing me with a chequeing account depend on the number of transactions I make, rather than the size of my balance.
So what we would expect to see in equilibrium would be banks offering chequeing accounts that pay the full rate of interest, but have fees based on the number of deposits and withdrawals.
AFAIK, they do offer some big accounts that look like that, but for most of us they don’t. It looks like they subsidise transactions by paying a lower than market rate of interest and charging lower transactions fees. I don’t know why they do this.
Here it is, Nick:
http://traderscrucible.com/2011/04/25/nick-rowe-begins-the-walk-to-mmt/
RSJ:
“Therefore banks will continue to make loans up until the marginal costs (e.g. FedFunds + capital requirements”
It should be capital requirements + liquidity cost, with FedFunds being just one of the liquidity sources. Liquidity cost is notoriously hard to model and price correctly, ALM procedures notwithstanding.
One is quite flabbergasted by some folks, who should know better, de-emphasizing the role demand deposits play in liquidity provision and going as far as to say that demand deposits are not even needed for bank operations. As a matter of fact, retail deposits are the most stable source of liquidity statistically speaking. Just ask Northern Rock who thought that wholesale market can provide all liquidity they need.
I went through that retail deposits subject several times in the past, with little success, so it’s my last message on the subject 🙂
K (or RSJ),
Can you explain the rationale to me–the govt is just as good at banking as the private sector (maybe even better), so the govt can take over whatever aspects of banking with no loss to society as a whole and the govt gets to eat the banks’ lunch? Is that right? (I’m thinking of RSJ’s option 1 in his post “Leaving (Modern) Money (Theory) On The Table”, and k’s preferred option).
Vimothy,
Yes, I think pretty much anyone can offer vanilla deposit services. This does not require talent.
Extending loans and credit analysis requires talent. But no one is talking about squeezing banks out of the lending business, but rather taxing the profits they earn from providing deposit services.
Some nations — e.g. Japan — has its post offices provide deposit services of this form. I would be surprised if it was harder to to do this than to deliver mail. But we are talking about huge cash-flows. If the spread between MZM Own Rate and FF is 2%, then 7 Trillion * 2% = 140 Billion per year.
There’s no way that this money is spent on the costs of retail banking.
Trouble is, if you tax the banks offering deposit services, some other financial institution will find a way around it. Or banks themselves will. I think (not sure) that was the final straw that drove Canada to zero required reserves and interest on reserves. Banks said it was unfair that only they were taxed.
Nick,
You can apply the same argument to any sort of tax, and to any sort of benefit.
Banks are highly regulated institutions. It’s not a question of difficulty of implementation, but of political will.
RSJ,
Do you envisage the govt making loans as well, or just taking deposits?
No, not making loans, just providing vanilla deposit services.
I’ve always that of that guaranteed $140 billion a year (in RSJ’s example) as the bribe we pay the banks to have them submit quietly to regulation. They get easy risk free money and we avoid a repeat of the ’30s. At least that’s the theory… In the US at least things didn’t quite turn out that way.
RSJ:
“If the spread between MZM Own Rate and FF is 2%, then 7 Trillion * 2% = 140 Billion per year. ”
Should not it be “were” rather than “is” ?
It is 0.15% today (0.25-0.1). Therefore, $1.1T * 0.15% = $1.65B.
Where’s $7T come from ?
Btw. you tax idea seems good, but the banks will find a way to pass it to the consumers, more likely than not 🙂
Related question: What are the returns on MMMF and MMA like?
Could be that any MZM-FF potential spread is about to disappear anyway…
Vimothy/VJK,
LOL, yes, in a zero rate environment, there isn’t much of a spread. I was thinking, say, a bit longer term. Perhaps I’m too hopeful?
P.S. MMMF and savings deposits are part of MZM. Time deposits are not. Not sure where you would get breakdowns for time deposits or for MZM components. The MZM Own Rate is a weighted average.
Thanks JKH. That’s the one.
RSJ: sure, you can make that argument about any tax. But remember one difference: the inflation tax is more distortionary than other taxes like VAT or income tax (it causes a bigger welfare loss triangle for a given rectangle of tax revenue). That’s because you can avoid the inflation tax by reducing your money income (just like income tax and VAT), but you can also avoid it by increasing velocity of circulation for a given money income.
I was referring to your 10:22 comment regarding evasion of the tax on financial institutions. Perhaps we were arguing at cross purposes.
vimothy: I didn’t say anything about “whatever aspects of banking.” In particular, I said nothing about lending, i.e. credit brokering. I’m merely proposing that the central bank let everyone have digital currency, an obvious 21st century extension of the paper money they already provide to us and the digital money they already provide to the commercial banks. Additionally I’d prefer that commercial banks raise capital in the free market, without benefit of public subsidy in the form of grossly discounted deposit insurance.
Nick: “So what we would expect to see in equilibrium would be banks offering chequeing accounts that pay the full rate of interest, but have fees based on the number of deposits and withdrawals. AFAIK, they do offer some big accounts that look like that”
But they don’t. Rates on chequing deposits are exactly equal zero. That includes the almost $500Bn of “chequable deposits” in the BOC’s weekly financial statistics report which doesn’t even bother to quote the rate (though it provides values for every other type of account: overnight savings deposits: 0.05%, overnight savings deposits over $100,000: 0.5%, neither of which have changed significantly in 10 years or more). This the the distinguishing negative feature of a chequing account, and the only reason to have an overnight savings account. I challenge you to go into any large Canadian bank and get an interest bearing chequing account. There is no such thing. An efficient equilibrium is exactly as you describe. In reality, you earn literally, exactly zero, whether you deposit one dollar or one billion, and whether rates are at 1% like now, or 20% like 30 years ago.
Maybe the reasons are regulatory: I’ve been explained that since demand deposits can be withdrawn at any instant, they can’t pay overnight interest. In the US, Glass-Steagal prohibited the payment of interest on demand deposits, a restriction that was only removed with Frank-Dodd last year. I suspect similar constraints in Canada, though it’s not easy to tell: Regulations are created both at the BOC and OSFI, so it’s a tough slog.
A tax on demand deposits has little distortionary effect since no price is set off of it (I don’t see any relationship to the inflation tax). The marginal cost of overnight lending is the interbank rate. The principal effect is likely to be that the rate on savings deposits would shoot up to attract depositors out of chequing accounts. So it redistributes
seignorage profits, some (demand deposits) to the government and some (savings deposits) to the depositors.
” In the US, Glass-Steagal prohibited the payment of interest on demand deposits, a restriction that was only removed with Frank-Dodd last year.”
In the US, the G-S prohibition was circumvented as early as in 1970 by introducing negotiable order of withdrawal (NOW) accounts. NOW accounts have unlimited check writing privileges and are no different for all practical purposes from demand accounts (with some “for profit”. limitations). Both NOW and demand accounts are classified as transactional accounts.
The trick to bypass G-S was having a formal bank right to demand a seven day notice. The right has never been exercised in practice.
So, in the US, the reason that NOW accounts pay tiny interest rate is simple — there is no incentive/comeptitive pressure for the bank to pay more than they do today.
Historically, NOWs were motivated by competition from the MMMFs.
I do not remember what the situation is in Canada — it’s been a while 🙂
For each individual bank, deposits are demand determined. Households lend deposits to the bank whenever they want for as long as they want. They are each lending money for a small period of time.
But FF is the rate charged for banks borrowing reserves whenever borrower wants and in whatever quantity they want.
What would be the market mechanism that would drive these two rates to be equal?
But in aggregate we know that there will be a system level of deposits that is more less a constant function of income, as the demand for deposits is relatively interest-inelastic, and are set by timing of expected future expenditures. If the household doesn’t believe it will need the money for near term expenditures, then they purchase a bond.
Therefore in aggregate, the banking system as a whole will receive funding at below the fedfunds rate, but its marginal costs of funds will always be fedfunds.
To fix this, you don’t need to tax deposits per se. It’s enough to tax the balance sheet at the policy rate, and then give credits for interest payments to creditors.
That would convert marginal costs into average costs.
I still don’t see the connection with inflation. We want to impose taxes on banks, and we want banks to pass these costs onto borrowers, because creating money should have incur costs. The higher the tax, the lower the inflation.
I agree with vjk that banks have basically quit making even a pretense of competing on deposit rates. The supply of deposits to each bank is simply insufficiently elastic as a function of the paid rate. I decided to do the work of digging up and plotting the historical Canadian bank deposit rates. Here they are. The graph shows the official “Bank Rate”, the “typical” rate on overnight savings deposits and the rate on deposits over $100K. The trend is pretty clear, but I’ll try to do more illustrative graphs soon (tomorrow night, if I have a chance). Over the past 30 years, as rates declined, the rate paid by the chartered banks on deposits declined even more. The overnight deposit rate was around 2% lower than the Bank Rate in the early 80s but the spread increased gradually until the deposit rate hit roughly 0 in the early 1993 where it basically stayed ever since, even as the Bank Rate went as high as 8%. Deposits over 100k follow a similar pattern, starting at around a 1% discount in 1990 and increasing to a 1.5% discount until the Bank Rate hit 2.25% in Feb 2002, after which the Bank Rate rose from 2.25% back up as high as 4.75%, but the deposit rate completely disconnects and just gradually trends downward to its current 0.25%.
It looks to me like there is a new oligarchic equilibrium at 0% deposit interest. I am not at all familiar with models of monopolistic pricing, but it seems to me that it would not be that hard to imagine an equilibrium in which raising the rate would cause a greater loss of profit margin than it would attract additional deposit especially if the Bank Rate is fairly low, so a change in the deposit rate reflects a larger fraction of the profit margin.
Oops. Forgot to mention: the rate for deposits over 100K doesn’t actually exist before 1990. I simply held the spread to the Bank Rate constant going back to 1980. Otherwise the google graph widget wouldn’t work. Ignore the yellow line from 1980 to 1990.
Think of a world where Government debt has two components: interest-bearing bonds and non-interest-bearing ‘currency’ (this could be dollar bills, or zero-interest deposits in a state-owned bank, or commercial zero-interest deposits in a banking system with 100% reserve requirements and no interest on reserves). Then the average interest rate on debt, r*, is a weighted average of zero and the bond rate r. Even if r is greater than the growth rate g, so the economy is dynamically efficient, the possibility of seigniorage on currency means that r* may be less than g and thus permit Ponzi finance. Clearly the more the Government can induce the private sector to hold its wealth as currency (and thus avoid paying interest) the greater the scope for such schemes.
In the UK pre-1970 (I’m not an expert on Canada or the US) banks did not pay interest on current accounts, although the profits from this arrangement accrued to the banks (via a cartel arrangement) rather than the Government (via taxes or zero interest on reserves). Banks also sometimes charged for current account transactions. The result was that a relatively small proportion of the population held chequing accounts, and currency was much more widely used. Such arrangements might appear to increase the currency/debt ratio, and hence allow extra seignorage. But they also meant that most workers were paid weekly in cash, and that both allowed them to economise on cash (reducing seignorage) and imposed substantial real costs. So any argument which suggests that Governments can increase seignorage revenue by eliminating interest on demand deposits (with a corresponding tax on bank profits) must recognise that such a policy would also change payment arrangements (as well as the fact that a privately-owned financial system would rapidly invent substitutes such as money-market mutual funds). And I doubt if you could uninvent the credit card, except in the context of a socialist system which prohibited privately-owned financial institutions.
In addition, even if the Government can reduce r* below g, the argument that it can finance unlimited deficits through debt creation is flawed unless the private sector is willing to hold a very large amount of debt relative to GDP. If r* is zero (an extreme case), the overall deficit and primary deficit are the same (since there is no debt interest to pay). If the Government runs a deficit (measured as a fraction of GDP) which is greater than the product of the growth rate of GDP and the current value of (debt/GDP), the debt/GDP ratio will rise. So, for example, assuming a growth rate of 3% and a debt/GDP ratio of 200%, the debt/GDP ratio will rise for any deficit greater than 6% of GDP. It is true that this increase in the debt ratio will (assuming the interest rate remains at zero) allow the sustainable deficit/GDP ratio to be larger in the future: but it really is unclear why individuals should wish to hold vast amounts of (illiquid) zero-interest debt if other assets are available. Since the effectiveness of all such schemes depends on the difference between r* and g, a more realistic assumption about the interest rate on bonds means that the required debt/GDP ratio becomes larger still.
There are two further points about debt and functional finance which seem to be worth making. Firstly, many of the problems with functional finance and cumulating deficits disappear if Governments can run deficits (thereby ensuring full employment) while simultaneously acquiring productive capital. In this case there is an asset and revenue stream which corresponds to the higher debt, and (assuming the investment pays off) no future tax liability. I suspect that it was this modest ‘socialisation of investment’ which many supporters of functional finance (including Keynes) wanted to see. Secondly, the argument that debt interest payments are merely transfers between debtors and creditors ignores the fact that the interest payments are in effect lump-sum, while the taxes are not. (Incidentally, Barro’s original argument for Ricardian equivalence also makes the lump-sum tax assumption). If taxes are not lump-sum, then this is a further reason for rejecting the Ponzi escape route from the long-run budget constraint: significant deficits are only feasible if the debt/GDP ratio is high, and this will impose large tax-induced distortions. (I accept that in theory you could have a huge debt/GDP ratio and raise ALL revenue via Ponzi finance, but I don’t think that a policy in which there are never any taxes is credible).
William: good comment. Agreed.
One minor addition: “Firstly, many of the problems with functional finance and cumulating deficits disappear if Governments can run deficits (thereby ensuring full employment) while simultaneously acquiring productive capital.”
Agreed. But if the government were simply acquiring the productive capital that private investors would otherwise have acquired anyway, then there is a “direct crowding out” of private investment, so the deficit spending doesn’t increase Aggregate Demand. Instead of the private firm building the factory and issuing bonds to finance it, the government builds the same factory and issues bonds to finance it. (Or the government buys ownership of the factory, and replaces private bonds with government bonds. Or, the government buys the private bonds and issues its own bonds to finance them, so it’s really just acting as a financial intermediary.)
First of all, can we agree that the Long-Run Budget Constraint, as defined here, has zero content for policy? It’s descriptive, not prescriptive. It says that a shift toward deficit in this period necessarily implies a future-period shift toward surplus of equal PV, but it says nothing about whether such a shift is desirable. In particular, it does not provide any basis for talking about a “fiscal crisis”, as the very next post on WCI does.
One of the most striking features of this debate, from the heterodox side, is that there’s this disconnect between the consensus position on theory (the government net financial position is always zero in the long run) and the consensus position on policy (policymakers should accept large costs in order to keep the government financial position within certain bounds), yet neither side seems to realize it. Seriously, Nick, how to do you reconcile this post with the following one?
Anyway, I think I can convince you that the LRBC does not provide any basis for taking the existing stock of debt into account when setting fiscal policy. First off all, the standard formulation that you give here is just wrong. It shouldn’t be government debt on the left side, but total government liabilities and assets. The important thing about this is it raises the question, how do we know the government is actually on its LRBC? note that strictly speaking, the constraint is an inequality. It’s supposed to be satisfied exactly for households only on the assumption that they have no desire to accumulate wealth for its own sake, but only to finance consumption. That’s not a very realistic assumption, but at least it’s households; for governments, there’s nothing equivalent. As far as I can tell, the only reason to assume that the government is on its LRBC is “just because.” Well, of course in a closed economy if the private sector is on its constraint then the government must be on its constraint too, but that just goes back to the assumption that household’s desired stock of financial wealth is zero. Drop this ad hoc and unrealistic assumption, and there’s no reason that an increased deficit today can’t correspond to increased financial wealth held by the private sector, rather than to increased taxes sometime in the future.
But my main argument is that there is a logical inconsistency in your position. Question: Are deviations in aggregate income due to fiscal policy strictly temporary, or do they have a permanent component? In the first case, if in the long run output always converges to potential, and potential output is unaffected by current output, then all fiscal policy can do is shift income between periods. But, it follows from this that a downward deviation in demand today must correspond to an upward deviation in demand in some future period. So we never need to worry about the problem of taxes being required to meet interest payments, because if we borrow just enough to get AD up to potential output in a current recession, then the taxes required to pay off the debt will always be just equal to the taxes we would charge in a future inflationary period to bring AD down to potential output. So we never need to know anything about the debt. Following the functional finance rule “if inflation and output are below target, make the deficit larger; if they’re above target, make the deficit smaller” will automatically ensure that the long-run budget constraint is satisfied.
That’s if there is no permanent component to the changes in output resulting from fiscal policy. If there is, we can no longer assume that the need for expansionary policy today implies a need for contractionary policy sometime in the future. If the output gap follows a unit root process, then it is true that following functional finance rules could lead to changes in the permanent changes in the stock of debt. However, in this case there is no longer a present value of future taxes and expenditures, independent of the decisions made in this period. We can no longer say that government fiscal choices must satisfy the long-run budget constraint, because the long-run budget constraint itself varies based on current fiscal choices.
Your argument is contradictory because it assumes both that there is no permanent component to fiscally-induced changes in output (so that we can speak of a single long-run budget constraint) and that there is a permanent component (so that following a pure stabilization rule can lead to long-run changes in the debt/GDP ratio.) To go back to the example of whether finding some overlooked debt should cause us to run a lower deficit in the current period — in a LRBC world the answer is No, because that additional stock of debt also tells us that future demand growth is stronger than we expected, and so future tax revenue will be higher. Now you might say this is silly, and I agree — but if it’s not true then the long-run budget constraint is not defined.