Winner of the New Statesman SPERI Prize in Political Economy 2016

Thursday 4 July 2013

Government debt, Inflation and Money

Do budget deficits cause inflation? Let me be a little more specific: does raising the level of debt and keeping it there when the economy is at full employment raise the price level? The conventional answer is: not if the central bank controls inflation. Sometimes economists say the same thing in a different way: not if the debt is not monetised. High debt may be problematic for other reasons (e.g. crowding out of private capital, default risk, increasing distortionary taxation), but not because it must cause inflation.

This post is about explaining this conventional view. The two ways of giving the answer reflect two different ways to describe the conventional view, and I think that tells us something interesting - although perhaps controversial - about the role of money.

In the textbooks, the conventional view starts by talking about the demand for the medium of exchange, money. The amount of money in the economy, it is assumed, is related to the amount of money created by the central bank, but not the amount of debt issued by the government. The demand for nominal money is proportional to the price level, which is what economists mean when they say people demand ‘real balances’. So, if the stock of nominal money does not change, neither can the price level. When economists talk about not monetising the debt, they mean that the central bank keeps nominal money fixed.

There are two elements in this argument. The first has to do with the relationship between the amount of money created by the central bank (‘base’ or ‘high powered’ money), and what the financial institutions that create money (private banks) can do. The second has to do with money demand, which is what I want to focus on first. To do so, imagine an economy where the only money is cash printed by the government.

It is trivial to show why there can be this tight and simple link between cash and the price level. The economic system is all about real variables: not just consumption and output, but also relative prices like real wages. Furthermore people want money to buy some real quantity of goods, so the demand function can be written as M/p = f(...) where (...) includes real output. So, if the price level only appears on the left hand side of this equation and nowhere else in the system of equations describing the economy, and the central bank controls the supply of cash M, then this will lock down the price level. This is the famous neutrality of money. Furthermore, the mechanism by which this lock down works is intuitive: if the central bank creates more cash, we have ‘too much money chasing too few goods’, so prices rise.

Once we allow private banks to create money, the story can get much more complicated. The textbooks try and short circuit this by teaching the money multiplier, which I think does a lot more harm than good. But we could just assume there is some mechanism by which the central bank can control the amount of money created by banks, and continue to tell our neutrality story.

So according to this conventional view there is no worry about government debt, as long as the central bank ignores debt when ‘fixing the money supply’. Whether it always will ignore debt, or whether it always can, is a separate issue for another post. The critical assumption I make here that allows me to avoid this issue is that the fiscal authority does adjust its taxes or spending to make the higher level of debt sustainable.

This story is missing a key ingredient, and to see why consider the following. Let all government debt be nominal (not indexed). Suppose that, just as there is a demand for real money, there is also a demand for a real quantity of government debt: B/p = g(....). The government, by cutting taxes for a period, raises the supply of nominal debt by a fixed amount. Suppose it keeps nominal debt at this level. In that case, using an argument analogous to the earlier one involving money, will the price level not increase, until the supply of real debt matches the demand for real debt? If so, higher government debt has raised the price level.

One argument here is to say that, as the government increases the nominal quantity of debt, the demand for debt also rises in step, so there is no need for prices to rise. This will happen if consumers are completely Ricardian, because they believe tax cuts today mean tax increases tomorrow, and they save to pay for those future tax increases by buying government debt. In this sense, the supply of government debt creates its own demand, so we do not need anything else, including the price level, to change.

Suppose, however, that this process is incomplete, perhaps because some consumers are credit constrained, and so spend rather than save their tax cut. Does that not mean prices will still have to rise a bit to match the increase in the supply of nominal bonds? However, if we still have a fixed nominal amount of money, then higher prices will raise the demand for money, giving us a contradiction. What squares this circle is that interest rates rise, which makes people economise on money, and also raises the demand for government debt without the need for prices to increase. So higher debt might raise interest rates, but it will not raise the price level if it is not monetised.

Now an interesting feature of this story is that we could cut out the stuff about money altogether. We could just talk about the supply and demand for nominal government debt, and how the demand for debt is positively related to interest rates and prices. If the government wants to borrow more, the demand for nominal bonds needs to rise, and this can happen either because interest rates rise or because the price level rises. If interest rates rise sufficiently there is no need for higher prices. Loanable funds vs liquidity preference and all that. [1]

It is a small additional step to just talk about the central bank controlling interest rates to fix the price level. Nowadays this is how many macroeconomists would explain why higher government debt does not raise prices: the central bank changes interest rates to make sure it does not. This explanation not only has the advantage of simplicity (we do not need to talk about the demand for money, or how the central bank controls its supply), but it also seems to match how central banks think.

Of course something about money is there in the background. When we talk about interest rates being varied to control inflation, and why therefore we can ignore the size of the stock of government debt as an influence on inflation, we are assuming that the central bank has the ability to control interest rates. This depends on the fact that the government can issue money, or more specifically that the central bank’s “liabilities happen to be used to define the unit of account”, to quote from the bible Michael Woodford’s Interest and Prices (p 37). So money is there, but like the impresario of a play, it does not need to appear on stage.

In terms of the question posed by the title, both ways of describing the conventional view (with or without money) end up with the same answer to the question about government debt and inflation, which is good. However I remain puzzled about one thing. Do those who still tell the story using money think that telling the story just with interest rates is equally valid, or in some way misleading? When, with Campbell Leith, I first started using ‘cashless’ models of the Woodford type, a frequent complaint was ‘where was money?’. To appease potential referees we occasionally put money in, even though this added nothing to the main points of the paper. Yet I think those asking the question thought we might be missing something more fundamental, but I never discovered what it was. I remain genuinely curious.     

[1] Recall that I am assuming full employment in all this. In a recession caused by people saving more, higher saving will raise the demand for bonds, so even if the supply of bonds also rises following budget deficits, interest rates or the price level could fall rather than rise.


  1. How does velocity figure into the story?

    1. When Simon writes that too much money *chases* too few goods, he is implying that velocity does matter.

      GDP = P * Y = M * V, or P = MV/Y. This determines the price level. Inflation is the change in price level, so a big increase in M can usually achieve that.

      What about the original question on government spending? Well, it increases Y, but can also increase M and V. So not clear. But if what the Treasury giveth, the Fed taketh away through decreasing M (and perhaps decreasing V as higher rates encourage saving), Simon's point that the central bank can keep money fixed (not monetize) and therefore control inflation seems valid.

      Eliminating M from the picture requires a central bank that can otherwise control V and real GDP (Y/M). If you grant it that power, then the conclusion should hold.

    2. And I should add that, from most economists' point of view, velocity doesn't matter as it is just an unstable variable defined by GDP/M!

  2. The discussion seems very similar to the old way of justifying the omission of explicitly showing the bond market in simple 2-dimensional IS-LM analysis. I.e. that Walras' Law allows omitting one market in a 3 market model.
    In the more modern version in this post, a similar conclusion seems to follow: for comparative statics (i.e. only worrying about the initial and final positions) it doesn't matter which asset you omit (money or bonds). If you want to tell a coherent economic story about the causal mechanisms that lead from the initial equilibrium towards the new one, then typically the account involves interactions in all 3 markets.

  3. "the governments wants to borrow more"... WHAT? What does the government want to borrow more of, if you've cut money out of your model?

    1. I suppose you'll say the answer is "nothing"? You don't really mean "the government wants to borrow more", you mean something like "the government decides to do a helicopter drop of bonds"?

      It seems like a form of sophistry. If every good, asset and service in the economy was priced in terms of government bonds, and government bonds were exchangeable only for other government bonds, then maybe we could talk about bonds being the medium of account. But, instead, we use money. I guess I've missed the real point.

    2. If bonds were
      * the medium of account
      * the unit of pricing/value
      * the safest store of value (absent what we currently colloquially call 'money')
      then bonds *would be* money in the functional sense. i.e. 'money is what money does' or rather 'money is how money is used'.

      The fact that these functions/uses currently happen to be most closely performed by the particular legally constructed financial instruments that we colloquially call 'money' (Reserves, cash, deposits liabilities of a regulated class of financial entities called Deposit-Taking Institutions in the UK) is the result of our particular institutional-social history.

  4. I’m puzzled by the idea that money “does not need to appear on stage”. Money plays a leading role throughout the performance, which goes something like this (I think).

    Government cuts taxes and borrows more. The private sector’s stock of money remains unchanged because money borrowed is spent back into the private sector. But the private sector gets additional bonds. I.e. private sector assets rise. That’s stimulatory. So government / central bank has to raise interest rates, which it does by offering to borrow at a higher rate than the going rate, which withdraws cash from the private sector. A new non-inflationary equilibrium is attained, and money has played a leading role throughout.

    Didn’t you write money out of the plot when you said “What squares this circle is that interest rates rise…”? I suggest that should have read “What square this circle is that the central bank withdraws money from the private sector by selling bonds that have a new and higher yield.”

  5. I guess I am confused by the question. In the case of an economy already at full capacity are you asking if government policy that increases peoples ability to spend via a tax cut would necessarily be inflationary? Wouldn't it always be inflationary unless you convinced people not to spend that tax break? Even if the central bank raised interest rates to try to cause people to save rather than spend wouldn't increased interest income be inflationary? Unless the savings were immediately used to invest in things that actually increased the economy's capacity to provide goods and services in the short term? Maybe I dont understand what causes inflation in the first place. Im going with demand for goods and services that increases faster than the ability to supply those goods and services. And sometimes a supply disruption of hard to substitute for goods I guess. What is your basic theory? Sorry for all the questions.

  6. The quantity view has particular details of the financial system embedded in it, for example, that base money doesn't pay a competitive rate of interest.

    The interest rate view abstracts away those details, and is therefore simpler and more general. No matter how the financial system evolves, it will always be right.

  7. I have to say that I am unsure of the point being made in this blog-post. But, if you were to put this question to me in an interview, here would be my attempt at answering it...

    "Does raising the level of debt and keeping it there when the economy is at full employment raise the price level? The conventional answer is: not if the central bank controls inflation".

    ... and that's assuming the CB controls inflation how? I would imagine that the CB controls inflation by being able to raise interest rates so that the new spending represented by the new debt being emitted (or the new tax cuts being given) are neutralised i.e the tax cuts are indeed saved or the gvt spending does crowd out some private spending.

    So I find the exercise a bit weird. Why would the fiscal authority try to make it harder for the monetary authority to control economic activity when we're already at full employment?

    Surely, the whole debate with debt and monetazing debt right now is that we're NOT at full employment/full capacity and that some people insists deficit spending, let alone monetized deficit spending, is still inflationary...

  8. I think lars christensen would say that you are confusing "money" with "spending", and that "spending" is the `real' thing.

    With output held fixed in the short run, an increase in the monetary base will only not cause inflation if velocity falls. This is essentially what you mean when you say "if consumers are perfectly ricardian", i.e. if total spending is held constant. It is perfectly possible for a CB to keep increasing the base but also keep raising interest rates, in such a way as to maintain spending/demand, and by the SRAS curve, that must maintain the price level. One is expansionary, the other contractionary, and so one can balance spending. Yet this possibility appears impossible in the woodfordian `real' framework. So yes there is something missing. - though I am not an expert on woodfords work.

    Another way of stating the above is to point out that, if QE is expansionary, and raises spending, as it appears to be, then it is expansionary regardless of the interest rate at which it is undertaken. Indeed, when the difference between cash and Gilts is larger, QE is more expansionary. So you can always do QE rather than cutting short term interest rates. (Not sure CB has much control over long term real interest rates - unsure, did you see the Glasner post?)

    How to understand this? Essentially, it is because of credit creation. If I hold the base money constant, but I increase the broad money supply through credit creation (e.g. issuing government debt) then I must have more spending somewhere. The only way I can offset this is by persuading someone some where to increase their demand to hold "broad money" i.e. if velocity falls. This is the story you are telling with your "ricardian effects". Gilts can be exchanged for money, therefore just like money, they have a "gilt demand".

    In practice, although there is a new equilibrium in which people choose to hold just the right amount of extra gilts, there is a period of higher spending before it is reached because Gilt demand is sticky just like money demand, which is why in your above story there would be a short term bounce in real life. The economy "feels its way" to the new equilibrium, producing a temporary bounce in spending for all the same reasons that increasing the base money would have an effect. Probably a smaller effect, since gilts are held mostly by financial institutions which are probably a lot less "sticky" than domestic money demand.

    So in reality, even if the ricardian equilibrium was the end result, you would get a rise in the price level followed by deflation, and end up at the same price level in "the long run".

    In essence, this provides the answer to whether CB's are responsible for the credit boom. If they target spending/inflation. A CB can do it by interest rates alone, or by controlling the money base alone, or by some combination. The difference between these strategies is how much of the spending is created through credit creation. I suspect that large scale QE along side raising interest rates will enforce widespread de-leveraging - exactly as appears to be happening in the US.

    PS: Coming from a background in theoretical physics, one of the first things that you learn in Quantum is that the thing that you measure is the "real thing". Thus dividing up spending into "real GDP" and "inflation" just seems intuitively wrong. Sure, they are an interesting sideshow, but in reality it is (expectations of) nominal (measurable) values which drive the economy, because the vast majority of economic actors do not ask "what is the real return on my project, and what is expected inflation", they ask "what is the nominal return on my project, is it higher than the nominal return on capital". If this is how economic actors behave, then it should be possible to construct agency models/theories without any "real" factors at all.

  9. Do those who still tell the story using money think that telling the story just with interest rates is equally valid, or in some way misleading?

    I suppose that if you assume monetary equilibrium (ie, the CB moves to keep spending fixed), then money does not make for a very interesting addition to the model. However, when monetary disequilibrium arises (say, the CB thinks its hands are tied because of the ZLB), money becomes an important feature of the model, because the CB can increase or decrease nominal cash balances at will. Given sticky prices, then short-term real cash balances are also controlled by the CB. Ergo, the CB is not powerless at the ZLB.

  10. Simon: good question.

    My (inadequate and unclear) answer:

    Yes, I do think you are missing something fundamental. But you are not alone. Anyone who writes "M/P=L(stuff)" as if money were just some other asset that people hold, like bonds, or houses, or cars (and that is nearly all of us, including a lot of "monetarists") is missing that same fundamental thing.

    Consider two (alternative) governments:

    1. This government can only issue very short term bonds to finance a deficit (some other entity issues money, or perhaps people use gold as money). The bonds offer a floating rate of interest Rb that the government can adjust. This government can *either* choose the deficit, *or* it can choose Rb, but not both. Because if it chooses the deficit, it must adjust Rb continuously so that people demand the flow of new bonds it needs to sell to finance that deficit. It cannot run a deficit unless it can persuade people to want to hold a growing stock of bonds.

    2. This government can only issue money to finance a deficit. The money pays interest Rm that the government can adjust. How many degrees of freedom does this government have? I say it has two degrees of freedom. It can choose the deficit, *and* it can choose Rm. Because money is the medium of exchange, people will accept new money in exchange for goods they sell the government even if they do not wish to *hold* any extra money. They plan to get rid of it, by using it to buy something else.

    You cannot see the difference between 1 and 2 simply by writing down:

    Deficit = dL(Rm,stuff)/dt

    But there is a difference.

    (I got this idea, initially, from David Laidler. It's also there in Yeager. Simply because money is the medium of exchange, the issuer of money can force the stock of money to increase in excess of the stock demanded in a way that the issuer of bonds can not force the stock of bonds to increase in excess of the stock demanded. The stock of money is supply-determined while the stock of bonds is determined by min{Qs,Qd}. This means that P is subsequently forced to adjust to ensure Qs=Qd for money in a way that it isn't subsequently forced to adjust for bonds.)

    I've done a couple of posts on this. None of them very clear, I'm afraid.

    1. One old post:

      A second old post:

  11. I'm a landlord, and I sometimes pay people with my own handwritten scrip that I promise to accept in payment of $1 of rent on my property. I never actually redeem scrip for central bank-issued dollars, just for rent on my property.

    People sometimes call my scrip 'fiat money', since it is not redeemable for gold, silver, or even dollars, but those people don't understand that my scrip is actually backed by my assets. If I want to mess with their heads, I tell them that I only issue scrip from my kitchen, which I call my central bank. If I really want to confuse them, I tell them about how I just issued 30 units of scrip from my kitchen and used it to pay off an old $30 debt of mine, thus conducting an open-market operation.

    So, to answer your question of whether budget deficits cause inflation. Of course they do. As soon as people in town hear that I am having trouble paying my bills, my scrip will lose value.


Unfortunately because of spam with embedded links (which then flag up warnings about the whole site on some browsers), I have to personally moderate all comments. As a result, your comment may not appear for some time. In addition, I cannot publish comments with links to websites because it takes too much time to check whether these sites are legitimate.