Recent Posts

Sunday, January 5, 2020

The Debt-To-GDP Ratio Does Not Matter: The Fiscal Constraint Debate

Most debates around Modern Monetary Theory (MMT) revolve around the role of operations and the meaning of the “fiscal constraint”: what are the limits on fiscal policy? (The focus on this one topic is either the result of this being the most interesting topic, and/or critics not being bothered to read anything else in the MMT literature.) These debates are typically uninteresting, because they end up being purely semantic debates, and the two sides actually agreeing on the underlying principles. To avoid wasting time on such debates, we need to re-cast the debate into a debate over something concrete. In the absence of a discussion of a specific policy, I believe that the best way to phrase the debate is as follows: does the government’s debt-to-GDP ratio matter?

There is a secondary, debate: do government’s need to fear “bond market vigilantes”? Since the usual argument is that said vigilantes will rebel when the debt-to-GDP ratio gets too high, it can be viewed as a side argument of the main debate.


Regular readers may have found a certain amount of fatigue in my writing about MMT debates. They have largely devolved into an extremely uninteresting form, where a critic argues that MMT implies something, MMTers respond that this is not the case, and then the critics complain that MMT writings are too vague, or that goalposts are being moved.

In order to avoid wasting time, we need to focus on concrete concepts. In the case of the financial constraints, the MMT literature will often say something to the effect that “there are no financial constraints on a currency sovereign’s fiscal policy (the United States Federal Government would be an example of a currency sovereign), only real constraints.” The crux of the issue: what is a “financial constraint”? Although I think the MMT literature explains the point clearly, obviously some people disagree.

My argument is that we can do a decent approximation of the concept “there is no financial constraint” by saying that “the debt-to-GDP ratio does not matter.” This is perhaps not enough to cover everything, but at the same time, I believe that it would eliminate almost all the fairly pathetic attempts to debunk MMT that I have run across in recent years.

An important point to note is that this is a negative statement: that the ratio does not matter. In order to contradict the MMT statement, the critic needs to come up with a story why the debt-to-GDP ratio does matter. The problem with the failed critiques is that they do not advance a similar argument, instead they make hyperbolic statements about fiscal policy that MMTers accept (e.g., really loose fiscal policy is inflationary).

Update: A key point is that I wrote debt-to-GDP ratio, and not the deficit. Since debt can be seen as related to past deficits, I have got the comment that they are related. In order to clarify further, imagine that you are given only the debt-to-GDP ratio for one time point, and that is all the information you have. Can you conclude anything based on that one data point?


I view the arguments here as an offshoot of my discussion of Warren Mosler’s MMT white paper. The reason is straightforward: they are my reflections on a verbal comment made to me by Mosler, which I remember as follows:
If statistical agencies did not publish the governmental debt-to-GDP ratio, would it really matter (for good macroeconomic analysis)?
I would note that Bill Mitchell has made similar statements (to the effect that the government should not target the deficit or debt-to-GDP ratio), which have influenced my thinking. I would note that I had independently come to the view that the debt-to-GDP ratio was not particularly interesting courtesy of covering the Japanese rates market for a few years.
  • One initial point is that the debt-to-GDP ratio is a bedrock data series for a lot of terrible analysis of fiscal policy. My argument is that the appearance of the debt-to-GDP ratio is a pretty good guide for the dividing line between good and terrible analysis.
  • The next point is that the discussion is for countries with currency sovereignty, which I prefer to define as sovereigns who do not have to worry about involuntary debt default. Which countries qualify for that status can be debated, but if one wants examples, the United States and Japan in 2020 are obviously currency sovereigns, while Greece is not (courtesy of being in the euro straitjacket).
  • Finally, one could come up with some uses for the debt-to-GDP ratio, but they are rely on transforming it to be equivalent to another variable. The most important example is that the fiscal deficit can be approximated by the change in the debt-to-GDP ratio (assuming we have some auxiliary variable), and the fiscal deficit can be used to approximate the stance of fiscal policy. As noted earlier, imagine that we only have access to the latest debt-to-GDP ratio value, and so cannot back out the deficit.

No Magic Tipping Points

Returning to the argument, MMT statements about the absence of a fiscal constraint on government can be translated to: there is no level of the debt-to-GDP ratio which has interesting predictive powers.

In particular, there are the following implications.
  • The debt-to-GDP ratio is not a useful metric for gauging the default risk of a floating currency sovereign.
  • The debt-to-GDP ratio has very little or no useful information for the determination of nominal interest rates.
  • The debt-to-GDP ratio has no predictive power for future growth or inflation (although we might be able to back out predictions for the debt-to-GDP ratio given inflation or growth rates).
  • It makes no sense for governments to set fiscal policy to move the debt-to-GDP ratio to any particular level.
The last point shows that there are real-world divergences between the MMT view and the conventional: the Maastricht criteria targets a 60% debt-to-GDP ratio, while steering the debt-to-GDP ratio is commonly discussed in budget documents. (The Maastricht criteria could be defended on the basis that it is a currency peg system.) (And once again, no cheating by relying on transforms on the debt-to-GDP ratio, in order to approximate the deficit.)

If we return to the concept of fiscal constraints, we see that the argument is that the debt-to-GDP ratio does not form a limit on fiscal policy. Instead, the MMT/Functional Finance is that the limits on fiscal policy show up in form of real constraints: shortages of inputs (labour, capital, natural resources) and/or inflation.* As soon as any objection to a lack of “fiscal constraint” veers to discussing any of these issues (“hyperinflation!”), they are agreeing with the MMT/Functional Finance framework, and so we are free to stop paying attention to whomever is ignorantly raising these objections as a critique of MMT.

Is MMT Correct?

Once we accept that the best way to phrase the division between MMT and conventional economics on the question of financial constraints, one reasonably would ask: are the MMTers correct? Well, since I am in the MMT camp, it is rather obvious what I think.

If someone wanted to convince me that the debt-to-GDP ratio matters, they first need to deal with the horrible example: Japan. To date, I have not seen anyone clear that rather low bar.

Derivatives of the Debt-to-GDP Ratio

If we know the change in debt and monetary base levels, as well as GDP, we can back out the fiscal deficit (modulo various national accounting rules that drive a wedge between the fiscal deficit and the change in governmental financial liabilities). As such, we can come up with an approximation of the fiscal deficit by looking at the change in the debt-to-GDP ratio. To what extent the fiscal deficit provides macroeconomic information, we can use the change in debt levels to proxy this. Although the fiscal deficit in dollar terms can be misleading, there is nothing in MMT that says that it completely lacks information in practice. As such, the debt-to-GDP ratio as a variable could be rescued via the back door, but we are back to playing semantic games.

The other angle of attack to save the debt-to-GDP ratio is to look at debt service. We can generate scenarios where the average interest rate on debt changes, and thus the more debt outstanding, the greater the effect on interest payments by the government. My argument is that once again, this is relying on the fiscal deficit, and is just another linguistic twist to try to pretend that the debt-to-GDP ratio by itself is relevant. Since the average interest rate on the governmental debt stock can vary extremely wildly – and can be negative! – we are just debating the effects of fiscal deficits.

Low Rates Justifying Loose Fiscal Policy?

There is one argument about fiscal policy that is somewhat related to the lack of importance of the debt-to-GDP ratio. Proponents of MMT argue that the level of interest rates should not be used to justify an expansive fiscal policy, whereas conventional economists have argued that the low risk-free rates in the aftermath of the Financial Crisis were an impetus towards large-scale infrastructure spending.

Since both sides agreed on policy (governments should have loosened fiscal policy in the aftermath of the Financial Crisis), this easily devolved into pointless semantic arguments. The real debate is whether financial considerations (like the level of interest rates) should be taken into account for infrastructure spending decisions; MMTers say “no.” Since interest expenditures are related to the debt-to-GDP ratio, this topic of debate can sort-of be covered by the suggested dividing line.

(The other way of disposing the argument from the MMT perspective is to point out that the government can lock nominal rates at 0%, and it ends the appeal to low interest rates by fiat. However, that shifts the terms of debate, as it requires a change to the policy framework.)

Bond Market Vigilantes?

An alternative formulation of fiscal constraints is to argue that if policy is “unsustainable” (whatever that means), participants in the bond market will refuse to roll over governmental debt, forcing a default. My argument is that this argument is too vague: what triggers this behaviour? One is very hard-pressed to find legitimate examples of such behaviour among developed floating currency sovereigns.

Although conventional economists tell campfire stories about bond market vigilantes, some post-Keynesians spin gothic horror stories where foreigners refuse to fund countries. Rather than forcing a default, we are to believe that the bid for a floating currency will be zero. Once again, there are not a whole lot of episodes where developed countries’ currencies have gone to zero, so this debate is also largely detached from reality.

In practice, believers in bond market (or currency market) vigilantes either point to a stock (the debt-to-GDP ratio) variable, or flow variables (fiscal deficit, current account deficit) as triggering the constraint. The stock variable is exactly what I point to as being the main dividing line, but we can augment this with a debate whether there is a magical dividing line in flow variables that trigger fiscal (or currency) calamities.

Otherwise, discussions about bond market vigilantes are inherently pointless. Like other MMTers, I argue that operating procedures ought to be adjusted so that involuntary debt default is impossible. Since these changes are so minor, the argument is that there is no practical difference from those institutional arrangement to the current ones. One is stuck arguing about highly implausible scenarios under current operating procedures, for which there are no good historical precedents.

Concluding Remarks

It is a revealed preference that many people like wasting time making semantic debates about economics. For those of us not so inclined, we need to find concrete points of division. Re-casting arguments about the financial constraint as a discussion of the importance of the debt-to-GDP ratio is one such tactic.


* Steve Roth questioned the use of "real" with respect to inflation. I am following standard usage in the MMT literature, which uses the split "financial" versus "real." Since inflation is not a financial constraint, it qualifies as "real." It might be better to use "non-financial constraint," but that is a phrasing one would not encounter in the literature.

(c) Brian Romanchuk 2020


  1. On the subject of bond vigilantes, Bill Mitchell wrote an article titled: "The bond vigilantes saddle up their Shetland ponies – apparently"

    In the CNBC report, we read that these “bond vigilantes” are “are no longer under ‘lock and key’ and are free to push yields higher”.

    The comments are attributed to a “Wall Street veteran” one “Ed Yardeni”, who apparently introduced the term “Bond vigilante” to the nomenclature that financial journalists love to use to make something appear important that is not.

    The article even came replete with a picture of three wild west types on horses, with guns drawn, presumably attacking some stagecoach or something, with evil intent obvious.

    Shaking in our boots we are!
    [end quote]

    On the subject of debt-to-GDP ratios, and in the other direction, this from a recent article I read:

    And that meaningless debt-to-GDP ratio, which is one of the world’s “worst.” For the U.S., it’s a bit above 100%. Don’t you wish it was more like Russia’s (14%) or Zimbabwe’s (21%)? Or does Mr. Yeargin prefer Guatemala’s ratio of 25% or Nigeria’s ratio of 30%?

    Or how about Japan’s ratio of 238%? Which economy and which inflation would you prefer, Japan’s or Zimbabwe’s?

    Here is a list of national debt/GDP ratios for countries. See if you can see a relationship between that useless ratio and the strength of a nation’s economy. Save your effort. There is no relationship. As we said, it is a useless ratio.
    [end quote]

    1. Thanks. I got some push back on Twitter that I was making these examples up. This article is aimed to be used an introductory section of a book, and I don't want it bogged down with documenting issues with other schools of thought; I could cover that later in the book.

  2. "The Debt-To-GDP Ratio Does Not Matter: The Fiscal Constraint Debate"

    Although I view the macro-economy through a somewhat different lens, I would agree with your analysis. Like you, I can take some constraint-unrelated information from a change in GDP/DEBT ratio.

    That said, thanks for writing another thought-provoking article.

  3. If debt/GDP doesn’t matter in the way you describe (I agree), then it must follow in the same sense that it must be the case that debt, deficit, and interest expense don’t matter in the same way.

    I don’t understand the intended meaning here:

    “To what extent the fiscal deficit provides macroeconomic information, we can use the change in debt levels to proxy this. Although the fiscal deficit in dollar terms can be misleading, there is nothing in MMT that says that it completely lacks information in practice. As such, the debt-to-GDP ratio as a variable could be rescued via the back door, but we are back to playing semantic games.”

    If the fiscal deficit doesn’t completely lack useful information, then the same should be said about debt/GDP, debt, and interest expense. I don’t know what the “semantic” thing is about here. This part isn’t semantics. Its analysis and logic. (Debt is cumulative deficits, etc.)

    One aspect of all this is that “wasted time” in dealing with responses to MMT might have something to do with the fact that MMT is less than perfect than explaining its position. I certainly think that. I understand the message, but the clearest explanation boils down to a proper interpretation of elementary accounting, particularly in the area of central and commercial banking, and government treasury operations.

    For example, the benchmark point about “affordability” could be made far more simply and effectively by avoiding the somewhat sloppy “consolidation view” that appears to be so important to the founding “paradigm” (the state “spends first” etc.) The actual real world deconsolidated case would be far more straightforward in explaining these things. It’s a split between income/expenditure accounting (Treasury spending and taxing) and flow of funds accounting (Treasury bank accounts and deficit finance).

    And the fact that inflation issues are captured only on the income/expenditure accounting side makes it obvious that the finance deficit/debt side “doesn’t matter” in the sense that you describe.

    1. I had not written this out explicitly, but my intended meaning was the following: can we use the variable as an input to a (reduced) model? Obviously, if we are obsessed with stock-flow consistency, we want the level of debt; but at the same time, if we did not have direct measurements (which we obviously do), we might be only able to approximate debt/GDP.

      Deficits are an income flow to the "non-government" sector, and it is highly plausible that it matters. Whether breaking out interest costs -- which is standard procedure -- adds value is questionable.

      If the only reason that the debt/GDP matters is because it is a transformed version of a variable that matters, it doesn't really matter.

      The consolidated view is the only one that makes sense from an analytic standpoint. Even if government accounts are presented on a deconsolidated basis, it is trivial to consolidate the data. Very few economic models have as many variables as appear in published national accounts, so we cannot pretend that models directly map to the national accounts.

      If we look at deconsolidated models, there is no information gained. They add equations that have zero influence on the model solution. Eliminating them reduces tractability issues, and makes the models easier to understand. The only conceivable reason to add the split is if one wants to model a default; but in that case, pretending that the central bank is anything other than a wholly-owned subsidiary of the fiscal arm of government is wildly misleading.

  4. From twitter:

    “Got a complaint on my website arguing that consolidation of the central bank and Treasury (fiscal arm) is confusing. How is possible to believe that adding equations that have no effect on the external solution simplifies models?”

    It’s not possible. My comment was not about modelling. It’s was about first principles in understanding monetary operations.

    I’m not particularly interested in modelling. Modelling is your wheelhouse, and it’s up to you as to whether consolidated or unconsolidated modelling techniques are more illuminating or efficient or effective.

    In fact, consolidation of balance sheet outcomes is a trivial exercise - in any institutional context - including this one with separate central bank and treasury operations. Consolidation is not only easy – it’s absolutely necessary in order to understand the joint effect of monetary and fiscal operations. This should be obvious.

    But it’s not sufficient in understanding the actual process of fiscal and monetary operations. There’s a difference between understanding the consolidated outcome of real world operations versus inventing a consolidated flow pattern that is only logically correct under different institutional arrangements.

    Where modelling is the concern, this doesn’t matter – because it’s the same balance sheet outcomes you’re interested in. But confusing the two doesn’t reflect the reality of what happens between outcomes.

    Most of your MMT critics are interested in how the real world process operates – more than model results. Understanding the facts of monetary architecture and process and explaining this without ambiguity or interpretational bias is different than “modelling”. It’s an education issue.

    For example, one negative outcome in this case is the number of critiques of MMT that are premised on the assumption that MMT is all about money financed deficits. This is a well-known pattern of misunderstanding over the three decades of MMT’s existence. MMTers understandably become vexed at such persistently false representations of what they intend to portray. Of course it’s wrong. This is seen easily by anybody who takes some time to investigate various writings on the subject.

    MMTers may want to reflect on this. MMT has itself to blame. The people that believe this about MMT certainly aren’t all stupid. Something else is wrong – and I don’t think it’s because critics don’t read enough MMT literature. It makes no sense that such widespread consistently erroneous interpretation results from something that is not seen.

    I think this particular misunderstanding relates directly to a rather weird portrayal of fiscal/monetary consolidation in flow dynamics (not balance sheet or model outcomes). It’s not a normal version of financial statement consolidation. The trouble lies in a strangely constructed interpretation of the interface between the central bank and the treasury function in the flow dynamics of payment relationships.

    … cont’d

  5. … cont’d

    The MMT notion that the central bank provides the reserves necessary for the private sector to pay taxes and that this somehow translates to “the state spends first” (or the occasional MMT comeback “spends or lends first”) presents logically challenged framing along multiple dimensions:

    # 1 – the conflation of GDP directed government spending (or even fiscal transfers) with central bank creation of reserves is dysfunctional from the perspective of elementary accounting vocabulary and comprehension. Spending is part of GDP expenditure/income accounting. Reserve flows are part of flow of funds accounting.

    # 2 – the MMT dominant focus on bank reserves is a source of deep trouble in the coherence of the “paradigm” and resulting education effectiveness. Coherent reference to the liability side of commercial banking is required as well. For example, the fact that bank reserves are necessary for tax payments is obviously not a sufficient condition that private sector taxes will be paid. And the provision of reserves for bond auctions cannot ABSOLUTELY guarantee all bonds will be taken up. (Only central bank purchases can provide such an ironclad guarantee). This is incomplete logic and yet it features front and center in the shout outs of the MMT paradigm (Mosler, Mitchell, Wray, and Kelton easily come to mind. Fullwiler seems a little more restrained, interestingly).

    # 3 – on a broader scale, it is a moot observation that the state provides the bank reserves to enable tax payments. MMT’s priority focus on bank reserves means that the starting point for its analysis recognizes the factual existence of a banking system with central and commercial components. This assumption of real world configuration is constructive. But the reference to the role of bank reserves and the assumption of a real world banking system obliges a complete analysis in that context. And that means this: reserves are provided not only to enable tax payments but to enable ALL private sector originated payments that cross ALL individual bank boundaries – central or commercial. And it is a fact that the real world system is designed to clear credits and debits for ALL transactions in a similar way – public and private sector. And it is a fact that this design intends that government Treasury payments are normally cleared in this way, with no net effect on the banking system money supply (complete netting via TGA and TTL mechanisms). This is because the design of the central/commercial system deliberately intends that net money creation be delegated to the commercial system. The reserve system is just an enabling device to facilitate this objective in a competitive commercial system. There is no reason to have bank reserves other than for a competitive commercial system to make loans and create money and compete in doing so. It exists entirely to enable competitive adjustment of balance sheets among commercial actors. It is of secondary functional importance relative to the assumed real world financial system design. (While that order can be distorted under conditions of emergency quantitative easing, that’s just a triage exercise aimed at fixing a cycle break in the commercial system.)

    MMT’s asymmetric portrayal of the relationship between the state and bank reserves is in logical conflict with the purpose of the reserve system in supporting broad functional requirements for commercial banking. The reserve-centric “MMT paradigm” focuses disproportionately on a system mechanism whose real world operation in fact is intended to neutralize such potential asymmetry.

    None of this matters for modelling balance sheet outcomes. But I wasn’t talking about modelling.

    1. Anonymous, what if we simplified your points enormously and just imagined a government that imposed and enforced taxes on all the citizens and just issued a paper currency that the citizens needed to pay their tax obligations in. And forget about the Central Bank and all other banks altogether because they don't exist. And all the stuff about reserves and who needs them and what they are for, well that's all gone as well.

      So the government could still spend more than it taxes and if we wanted to, we could call that a 'deficit'. And we could call the sum of the yearly deficits the 'government debt' if we wanted to do that also. And we could compare that to total economic activity and call it a 'Debt-to-GDP Ratio' and Brian's post would still be correct that that ratio doesn't necessarily tell us anything important about anything. What might tell us something important is an estimate of how much of that paper currency the public desires to save rather than spend and maybe that could give a guess to its future value.

    2. I’m tied up with other things, so only had time to skim.

      - MMTers argue that the central bank is forced to do the bidding of the fiscal arm. There is no choice but to coordinate operations. The narrative about “central bank independence” is wishful neoliberal thinking. Since operations have to be coordinated, there is no point in drawing arcane divisions between the central bank and the Treasury.
      - MMTers have done tons of research on the implementation of monetary operations. Way more than I am interested in reading. If you have a beef with that literature, I’m not the person to talk to. Everything I looked at seemed legit.
      - The “government spending first” narrative is something I have no opinion on. I don’t normally use it, and I wrote an article throwing my hands up in the air about it.
      - I wrote up operations in “Understanding Government Finance” from a “no reserves” base case, which I think is the best way to approach the topic,
      - I have no idea where reserve creation is being conflated with spending. I cannot think of any reference that is incompatible with national accounting.

  6. This comment has been removed by a blog administrator.

  7. Jerry Brown

    I appreciate your comment.

    I’m familiar with the idea of financial statement consolidation from a number of perspectives. As I said above, it’s a very useful device, and especially so in the case of understanding the joint effect of a central bank and a government treasury function. It’s an essential perspective in this case.

    Financial statement consolidation is one thing. It’s a legitimate and necessary tool for financial analysis.

    The MMT version of “consolidation” is quite another. It’s an utter mangling of the normal concept. It is a completely dysfunctional analytic approach to the subject of fiscal and monetary policy, and an impediment to effective learning about how the system works in fact. No wonder so many intelligent observers are still confused about what MMT is trying to say after 30 years. The message is thin gruel for mass consumption – enthusiastic supporters tend to shout out nutty half-baked monetary mantras as if at a Donald Trump rally for monetary theory.

    Again, here is the problem:

    The MMT “paradigm” is the original Mosler “paradigm”. The founder’s “paradigm” at its conceptual core is bank reserve-centric in the extreme. It is fundamentally unbalanced in this way. Because of this core perspective, MMT (unintentionally or unknowingly) implicitly invokes the full real world platform of central and commercial banking – without realizing the complete logical implications of doing so. That real world platform is set up in such a way as to fully integrate the operations of Treasury and the Central Bank with the operation of the commercial banking system. That means an explicit delegation of responsibilities to each those players in the system – including the delegation of the most important and most relevant primary money creation responsibilities to commercial banking.

    The notion that there is a sensible consolidated balance sheet outcome to the ongoing operation of that system is part of the logical design of the full architecture. And any decent analysis of what’s going on in that real world system needs to be checked for completeness of analytic technique.

    The MMT paradigm mangles that notion of logical completeness. It cherry picks certain parts of the system in isolation, and spins a tale that is biased toward the ideology of state dominance – utterly contradicting the inherent delegation of responsibility to commercial banking competition in that real world bank reserve platform.

    In that context, the notion that “the state” creates money every time it spends or that “the state” creates the only thing that is necessary to pay taxes is hogwash. Such a claim ignores the integration of “the state” with the commercial banking system, ignores the intended design of the bank clearing system in the context of that integration, ignores the liability side of commercial banking (the private sector source of tax payments), and ignores the same temporary “money creation” phenomenon in the case of intra-commercial bank clearings – i.e. it ignores the intended symmetric design of state and private sector credit/debit netting within that system. The MMT distortion of systematically cherry picking one side or the other of the reserve crediting and debiting process in standard payment netting (e.g. "creates money by spending") is a finagled interpretation of how the system works in the real world.

    The MMT version of “consolidation” is a convenient fairy tale for the targeted masses but an utter distortion of the full truth of the matter.

    1. The only hogwash I see is what you just wrote. MMTers have a lot to say about the banking system, and how it interacts with the government.

      Everyone who aims to be understood is forced to simplify matters, and focus on core relationships. Details are then added later. If you tell me the real world is complicated - so what? I know that. We need to abstract away complexities in order to grasp what matters.

      You have written a wall of text. I see no content that helps me understand anything, nor where MMT is wrong. If I wanted to explain what is happening in the banking system, I am not seeing anything I could use. The only useful content is that banks exist. I assume that my readers noticed that. Beyond that, so what?

  8. This comment has been removed by a blog administrator.

  9. This comment has been removed by a blog administrator.

  10. This comment has been removed by a blog administrator.

  11. This comment has been removed by a blog administrator.


Note: Posts are manually moderated, with a varying delay. Some disappear.

The comment section here is largely dead. My Substack or Twitter are better places to have a conversation.

Given that this is largely a backup way to reach me, I am going to reject posts that annoy me. Please post lengthy essays elsewhere.