What does it mean to “finance government spending by purchasing bonds”?

Here’s the Financial Times:

Torsten Sløk, a Deutsche Bank economist, pointed out the US central bank is not only under pressure from the White House, but also from progressive economists and lawmakers, where there is a growing belief that central banks can help finance extra public spending by purchasing government bonds, without triggering dangerous levels of inflation.

I see this sort of thing quite often.  It’s not exactly wrong, but it’s so misleading as to be the moral equivalent of wrong.

Debt can be monetized in two ways, by using newly printed currency to purchase T-bonds and by creating zero interest reserve accounts at the Fed and exchanging them for Treasury bonds in an environment where market interest rates are positive.  In contrast, creating positive interest bearing reserve accounts does not provide revenue for the government, it merely swaps one debt for another.

As a practical matter, in a world of positive interest rates about 98% of debt monetization is cash for bonds, and 2% is the creation of non-interest bearing reserve accounts at the Fed.  So as a first approximation, when people talk about financing spending by monetizing debt they are talking about buying Treasury debt with currency.

If you are willing to tolerate hyperinflation, then you can use money printing to finance spending up to roughly 3% or 4% of GDP, at most.  If you are not willing to tolerate hyperinflation (and we are not willing to), then the amount of spending that can be financed by printing currency is trivial.

If bonds pay positive interest rates, then banks don’t want to hold large quantities of zero interest reserves.  If the yield on T-bonds is zero then banks are willing to hold large reserve balances, but in that case swapping zero interest base money for zero interest bonds doesn’t finance anything.

So yes, you can finance spending by printing boatloads of currency when interest rates are positive and create high inflation.  That’s a tax on money holders, which is every bit as unpopular as any other tax.

I wish people would stop talking about this option; it’s simply not important in the modern world.



24 Responses to “What does it mean to “finance government spending by purchasing bonds”?”

  1. Gravatar of Dave Schuler Dave Schuler
    21. May 2019 at 13:08

    You wrote:

    “If you are willing to tolerate hyperinflation, then you can use money printing to finance spending up to roughly 3% or 4% of GDP, at most. If you are not willing to tolerate hyperinflation (and we are not willing to), then the amount of spending that can be financed by printing currency is trivial.”

    Is that formulation what you intended to write and is it correct? The 2018 deficit as a percentage of GDP was 3.8% and it’s estimated it will be around 5% in FY 2019. It didn’t produce hyperinflation.

  2. Gravatar of ssumner ssumner
    21. May 2019 at 13:25

    Dave, Yes, those numbers are correct. But deficit spending is far less inflationary than money printing.

  3. Gravatar of jim bowerman jim bowerman
    21. May 2019 at 14:06

    why is 3-4% that max that can be financed? Couldn’t we in theory have government expenses that run at 15% of GDP and which are fully paid for with newly printed money? This wouldn’t be hyperinflationary, would it? inflation would be 15% a year or so, no? quite high yes, but not hyperinflationary

  4. Gravatar of ssumner ssumner
    21. May 2019 at 14:42

    Jim, No, the peak of the Laffer curve for money creation is estimated at around 3% – 4%

  5. Gravatar of jim jim
    21. May 2019 at 15:03

    I think i’m wrong here, but i’m just imagining a very simple economy with only a treasury (no central bank etc). Government spending is 15% of GDP, and (overly simplistic) 15 out of 100 workers work for the government. Why couldn’t the treasury simply pay the salaries of those 15 government workers directly with newly printed money every year? This would lead to high (15% or so) inflation rates, but not runaway inflation or hyperinflation, no?

  6. Gravatar of ssumner ssumner
    21. May 2019 at 15:57

    Jim, No, that’s wrong, you are ignoring the impact of that policy on real money demand. I’d suggest looking at the studies of maximum seignorage during hyperinflation.

  7. Gravatar of dtoh dtoh
    21. May 2019 at 18:01


    You said….

    “I wish people would stop talking about this option; it’s simply not important in the modern world.” ….

    Then stop talking about it. Stupid is contagious – you might infect your loyal readers.

  8. Gravatar of Benjamin Cole Benjamin Cole
    21. May 2019 at 18:33


    Michael Woodford is a smart guy. He says QE in combination with federal deficits is a helicopter drop. That sounds like money-printing to me.

    The Bank of Japan has been buying sovereign bonds (an amount equal to 100% of GDP) and running national deficits for a long time. They face possible deflation. Did the Fed’s QE program have an effect on inflation?

    I do not think I am stupid, and I don’t get it. Maybe I am stupid.

    Ray Dalio says money-financed fiscal programs are inevitable.

    Through research and practice, I have became skilled at planting fruit-bearing trees and bushes. I have less luck in monetary policy.

  9. Gravatar of Benjamin Cole Benjamin Cole
    21. May 2019 at 19:09

    Add on re QE=money printing.

    In Los Angeles (probably in the whole state of California) temples and churches can run bingo games, and collect profits, but can only use proceeds gained in their “capital budgets.”

    In other words, a rabbi, priest, monk etc. can haul in $1000 from the Thursday night bingo game, but must spend the lucre only on bricks-and-mortar.

    But waaaaaay back in the 1970s, I was taught “money is fungible.”

    So, of course, a temple, church etc. can run the bingo game and put the money in the capital budget. But, as any heretic can deduce, money previously allocated to the capital budget is instead allocated to salaries.

    So, are the bingo games paying for salaries, or not?

    This reminds me of the way some macroeconomists look at QE and federal deficits, and then say that QE was not money printed to finance the federal deficit. Not a helicopter drop.

    Is money fungible? Or not?

  10. Gravatar of Michael Rulle Michael Rulle
    21. May 2019 at 20:10

    Benjamin Cole

    Dalio has been waiting, if not hoping, for massive inflation forever. It never comes. That’s what he calls Inevitable. On the other hand, if one of these gonzo idiot socialists gets elected, their Fed picks will make Cain seem like Friedman and Scot’s car will never stop accelerating.

  11. Gravatar of Matthew Waters Matthew Waters
    21. May 2019 at 20:23


    If you look at changes in currency in circulation versus federal spending, it has only been around 2%. Interest on reserves is a real interest cost paid by the Federal Reserve, though the interest it receives on Treasuries.

    Helicopter money, as envisioned by Bernanke, is not just more money but more Treasuries. The Treasury would issue T-bills straight to the Fed and the Fed could in fact decide not to roll them over. The money is destroyed at the next Treasury auction.

    Increases to currency in circulation, even after 2008, has supplied only about 2% of federal expenditure.

  12. Gravatar of Market Fiscalist Market Fiscalist
    21. May 2019 at 23:13


    I think that in your model for every dollar that the government creates by paying the salary (by money printing) of the government workers this will have an impact on GDP both because of the Keynesian multiplier and (assuming the banking system is not mandated to be 100% reserve) the money multiplier by which the banking system will increase the money supply based (as Scott suggests) on demand to hold money.

    Any increase in base money due to government “money printing” will have a much greater impact on NGDP due to these two multipliers.

  13. Gravatar of dtoh dtoh
    21. May 2019 at 23:14

    OK this is getting too frustrating. Everybody has an overly complicated model in their minds. Until you simplify the model to its fundamentals you can not understand monetary economics.

    There are two boxes in the Model

    Box 1: The Banking Sector – The FED and Commercial Financial Institutions

    Box 2: The Non-Banking Sector – Individuals, Firms, and Government (the government simply being individuals acting collectively)

    When the Banking Sector provides (or is expected to provide) more money/credit to the Non-Banking Sector, NGDP goes up. When it provides less, NGDP goes down.

    That’s it. Everything else is just an insignificant historical artefact of how the model has been implemented.

    You can construct a theoretical model with the Fed and the Government in the same box, one with individuals and their government is separate boxes, or one with the Fed and Commercial Banks in different boxes, but this only serves to obfuscate something which is otherwise trivial and to provide job security to economists and rhetorical fodder to moronic politicians.

  14. Gravatar of Benjamin Cole Benjamin Cole
    22. May 2019 at 05:03

    Mathew Waters–

    Right. I was not talking about paper money in circulation. (Ken Rogoff is concerned that paper money in circulation will undermine negative interest rates. That is another topic. To facilitate the negative-interest rate regime, we may have to live in the panopticon. No cash, meaning every paycheck and transaction can be monitored, along with your e-mails, texts and phone calls).

    Okay, back on topic: When the Fed does QE, it digitizes up a few trillion dollars, and buys Treasuries from the 22 primary dealers, who deposit the digitized cash into their commercial bank accounts. The banks sit on the digitized reserves, “earning” digitized interest from the Fed.

    But the 22 primary dealers first buy the Treasuries from customers or on the open market before selling to the Fed. Curiously, no one knows what the Treasury-sellers do with their proceeds. AFAIK, the Fed has never surveyed this key fact.

    Suppose the Treasury sellers are spending their proceeds on cars, women and wine? Women can go through a few trillion rather easily. Then, thanks to women, you would have a boost to GDP. But if the Treasury sellers just re-invest, then you have asset appreciation (maybe—in globalized capital markets with hundreds of trillions of dollars of assets (think property), what does the actions of a lone central bank amount to? A bucket of water thrown against the tide?)

    But back on the federal government and money being fungible: Suppose the federal government sells $1 trillion in Treasuries to help finance the FY 2021 budget, but then also Fed buys $1 trillion in Treasuries with freshly digitized cash.

    This is not a QE-financed helicopter drop?

    Side story: Seems to me, due to the existence of globalized asset markets with hundreds of trillions of dollars of value, a lone nation can monetize its national debt without consequence. Kind of sneaky, but doable. The Fed bought back $4 trillion of Treasuries, and what happened? Nothing! Evidently, the cash the Fed digitized up was just swallowed by global capital markets with barely a burp. So the bond-sellers were made whole, and now the federal government owes money to itself. (Yes, I know the Fed is not technically part of the federal government. but it might as well be).

  15. Gravatar of jim jim
    22. May 2019 at 06:08

    @Market Fiscalist

    Begun reading up on maximum seignorage during hyperinflation as Scott suggested and still know he’s right but can’t shake a few things in my head (helpful link below – i’m still trying to understand all of it –


    Definitely agree that velocity would go up initially if we implemented the 15% scenario tomorrow, but wouldn’t velocity and money demand eventually stabilize if the market truly believed that it would be 15% every year forever?

    Stated differently couldn’t we have milton friedmans k% rule where k = 15%? If money printing was entirely used to pay for govt employees salaries, would their real salary go down over time? Govt increases govt workers salaries 15% every year in nominal turns but what would their salary raises be in real terms in the long run? much lower than 15% i’d guess…though as Scott has mentioned before, over the long run MB and NGDP are correlated so how can a 15% annual rise in MB over say 100 years result in anything other than about a 15% annual rise in NGDP over that same 100 years?

    Even better, couldn’t we set an NGDP target at say 17% (15% inflation + 2% real GDP)? Not ideal but wouldn’t be runaway inflation either?

    Sorry for all the basic questions….still working through this stuff in my head

  16. Gravatar of Market Fiscalist Market Fiscalist
    22. May 2019 at 06:50


    Lets say that the only thing the govt spends money on is its employees salaries and this initially is 10% of GDP and paid for by taxes. Assume also that the velocity of money is 10 and stays constant and there is 0% growth.

    If the government increases salaries for government workers by 15% by printing new money then this represents a 1.5% increase in money supply and (as velocity is held constant) also a 1.5% increase in GDP and (as there is no growth) 1.5% inflation. So govt salaries go up by 13.5% (15% – 1.5%).

    However after this pay par rise govt salaries are now greater than 10% of total GDP. So if if the govt continues to give 15% pay rises without increasing taxes then this will increase the money supply by an increasing amount each year leading to ever higher inflation and a falling real value of the pay increase.

    Eventually I think this model will hit a steady state with inflation at 15% and the real value of the pay increase at 0% and the size of the government salaries as a % of GDP no longer growing.

  17. Gravatar of Matthew McOsker Matthew McOsker
    22. May 2019 at 07:01

    TO me an important thing to keep in mind is that when the government deficit spends it issues reserves (spending), and the bond simultaneous drains that amount. So if the fed does nothing, it is a transaction that is “somewhat neutral”. The fed can then change the reserve/bond balance after the fact with OMO.

  18. Gravatar of Scott Sumner Scott Sumner
    22. May 2019 at 07:01

    Jim, If you print money at a hyperinflationary pace, then demand for currency will fall to roughly 1% of GDP, or less. At that level of money demand, even doubling the money supply each year only yields one percent of GDP in real revenue. As the pace of money creation increases, the real demand for money falls. That’s standard Laffer curve economics. People are not willing to hold very much of an asset that is rapidly losing value.

  19. Gravatar of Jim Jim
    22. May 2019 at 08:42


    Yeah that makes sense but what happens after demand for currency falls to 1% of GDP? Doesn’t it eventually reach a steady state as % of GDP?

    Yeah i’m prob making a mistake but tried to model it out in a spreadsheet below. I’m prob making an error somewhere in my calcs but can’t find it right now. any ideas?


  20. Gravatar of Paul Paul
    22. May 2019 at 10:48

    The fact that central banks engage in monetary policy by purchasing government bonds anyway shows how trivial the supposed difference between fiscal and monetary policy is, anyway. Central banks require the gov to issue debt in order to engage in open market operations.

    You could change OMOs to mean having the CB purchase corporate bonds, stocks, real estate, or something else, but then that becomes fiscal policy.

  21. Gravatar of ssumner ssumner
    23. May 2019 at 09:23

    Jim, No steady state is reached. You need to think about the Laffer curve. Real flesh and blood people are paying the inflation tax. Would you pay 15% of your income to the government in a “voluntary” tax by holding lots of cash during hyperinflation? Of course not. You’d switch from Zimbabwe to US dollars.

    Paul. Wrong. Fiscal stimulus increases the national debt, monetary stimulus does not. BIG DIFFERENCE.

  22. Gravatar of Tuharsky Tuharsky
    23. May 2019 at 21:15


    Has it ever occurred to you that monetary policy can finance spending, even if we are not willing to tolerate hyperinflation, not by being overly easy, but rather by being overly tight? Consider the case of Japan. Who thinks that the Japanese government could maintain their current level of government debt at no significant cost if they faced interest rates comparable to those paid by governments who maintained adequately loose monetary policy, say, Canada or Australia? A central bank is essentially a partial government enforced monopoly on the issuance of money. (I say a partial monopoly because commercial banks may issue certain forms of money as well, subject to reserve requirements.) Every other monopoly earns monopoly rents by tightening supply, not by loosening it. Why would the issuance of money be any different? By restricting the supply of money, a government may force the private sector to compete on a small supply of regulator approved “safe assets” – i.e, government bonds – thereby bidding yields on those bonds lower. It is not a coincidence that in crises caused by overly tight monetary policy the yield spread between such government bonds and corporate bonds tends to widen. Whether consciously or not – and in some circumstances, I think it is certainly conscious – central banks may allow governments to earn monopoly rents by tightening monetary policy, forcing capital into government bonds, with the side effect of starving the private sector of capital, thereby triggering a recession. If I was a government with a big budged deficit or a weak currency, the last thing I’d want banks to do is to lend to the private sector and thereby create more money. If the only way to prevent them from doing so is to effectively force the nationalization of the private financial sector by bankrupting it through the use of monetary policy, so be it.

    Also a slight nitpick on your response to Jim. Monetary stimulus CAN effectively increase the national debt IF the asset purchased is anything but existing government bonds. When the Swiss National Bank buys Euros or stocks, certainly you would agree it is effectively increasing the government debt, as the government would be required to buy back any money it issued if inflation became excessive. I bring this up only because I think it is important to note that monetary policy is not just “tight” or “loose.” It also matters what assets are on the central bank’s balance sheet, and what is being bought or sold to tighten or loosen policy.

  23. Gravatar of ssumner ssumner
    25. May 2019 at 12:39

    Tuharsky, I think I understand your argument, but it doesn’t seem to hold up. Abe eased policy slightly, and inflation rose from slightly below zero to roughly 1% since 2013. Yet nominal interest rates stayed at zero. Implying that monetary ease reduced the debt burden.

    I’m not sure if your monopoly rents model fits here. Central banks tax base money by issuing new base money, and hyperinflation studies find that the revenue maximizing hyperinflation rate is quite high, hundreds of a percent.

    If a central bank buys risky assets, there is still no first order effect on net indebtedness. Yes, there is more risk of a loss, which can have fiscal effects. But that’s much different from spending or tax cuts that moves the net debt up dollar for dollar. Also, governments can invest with a long time horizon, and usually come out ahead with risky assets. Indeed Switzerland and Singapore might be better off in the long run issuing even more public debt, and buying up more global stocks.

  24. Gravatar of Paul Paul
    28. May 2019 at 10:50

    Scott, CBs don’t have to purchase short-term government debt to engage in OMOs , but they currently do (or long-term debt in the case of QE). Explain how they could do that if government debt was not growing over time.

Leave a Reply