Recent Posts

Sunday, March 18, 2018

Understanding Fiscal Sustainability Debates

I have encountered a number of discussions of fiscal sustainability over the past weeks. In particular, there have been debates between proponents of Modern Monetary Theory (MMT) and mainstream economists. This article does not attempt to settle the debate (although I am in the MMT camp, and obviously biased), rather frame the discussion. One of the problems with the debate is that the sides tend to talk past each other, as they have a quite different theoretical views, and this article explains why.

I have written on this subject in the past, and I will be attacking it again in the coming weeks. I will be writing follow up articles to "The Curious Profit Accounting Of DSGE Models" which discusses the business sector cash flow analysis in a basic Dynamic Stochastic General Equilibrium (DSGE) model. The follow up articles have the working titles: "The Curious Household Accounting of DSGE Models," and "The Curious Government Accounting of DGSE Models." (These articles are essentially a re-write of arguments I made in earlier articles.) Since the models I am looking at have three sectors, one may note that this trilogy will cover all three sectors. Those articles will explain why I am skeptical about the mainstream approach. However, I am putting that skepticism aside for this article, and instead just discuss what the mainstream wants to accomplish theoretically.

To start, I am going to assert that when we look at questions like fiscal sustainability, we need to keep in mind trade-offs. This reflects my engineering education: there is not a single correct answer to a problem, rather we need to see what the advantages and the disadvantages of a particular course of action versus alternatives. This viewpoint is in contrast to a more binary view, such as: the government is bankrupt, and the currency is going to zero! Although such binary viewpoints are entertaining, they do not help judge how to set policy.

In this case, the argument by Modern Monetary Theory is that the trade-off is between running fiscal deficits to achieve policy goals versus inflation. The mainstream side has a difficult time with that argument. Although it ends up being similar to mainstream economists' views most of the time, there is often a desire to insert a binary notion of  "sustainability" into the discussion. That is, we reach a point where there is no trade off to discuss, instead we have a "unsustainable policy" or a "fiscal crisis." The rest of this article discusses how that binary view departs from the MMT world view.

Partial Models

Before getting into the debate between MMT and mainstream approaches, we will have an excursion into what I term "partial models." These models are widely used, by both heterodox and mainstream economists. It is likely that most long-term "fiscal sustainability" models fall under this classification.

These models can have a wide range of complexity. The most minimal versions just use a few variables to predict the next period's (typically annual) government debt-to-GDP ratio. If we can forecast accurately the primary deficit (fiscal deficit less interest), the average interest cost, and nominal GDP growth, we can pin down the next period's debt-to-GDP ratio (given the known stock of debt) accurately using the one-period accounting identity. The objective of these models is to extend the range of this forecasting exercise.

Over a short-term horizon, it is hard to object strenuously to such models. In fact, I would probably build such a model if I wanted to forecast the debt-to-GDP ratio over the next five years (presumably because someone was paying me to do so).

The problem is that these models have theoretical inconsistencies when compared to likely behaviour of sectors. What these models lack is systematic behavioural relations between the economic variables in the model. The lack of such feedback is the criterion of including a particular model in the class of "partial models." There might be some feedback relationships -- such as "dynamic scoring" in the U.S. Congressional Budget Office (CBO) model, but the coverage is only partial.

The lack of feedback allows different variables to have different trend growth rates. The inevitable result of long-term extrapolation of these variables is that they either go to zero, or "go to infinity" (technically, have no upper bound) as we lengthen the time frame. If someone hands you a partial model and the debt-to-GDP ratio goes to infinity, that's actually a defect of the model, not government policy.

It is straightforward to see that a trajectory implying a debt-to-GDP ratio going to infinity is implausible. I am assuming that we are discussing a floating currency sovereign that is following some form of inflation target. We make the further assumption that the inflation objective is met, and that real GDP growth reverts to some form of trend value. The net result is that nominal GDP growth will follow a constant trend in the trajectory, and there is no reason for the central bank to change the level of interest rates. Since bond yields are driven by rate expectations, the average interest rate would be stable.

(A breakdown in the bond market would break this assumption. One can argue that is what happened in Greece. I will put aside that discussion for reasons of space.)

Let us assume that the trend nominal GDP growth rate is 5%. A positive growth rate implies that the fiscal deficit has to be constantly widening. At a 100% debt-to-GDP ratio, the fiscal deficit has to be greater than 5% of GDP in order for the debt-to-GDP ratio to rise (since the denominator of the ratio is growing by 5%). This widens; it needs to be greater than 10% if the debt-to-GDP ratio hits 200%.

We are then stuck with a very awkward situation: the model is assuming that nominal GDP growth is unchanged at 5% of GDP, yet the government is running ever-increasing deficits. Those deficits represent income to the private sector, and so the implication is that there are parts of the private sector with very large incomes and financial asset holdings. Sooner or later, those holders of government liabilities are going to go on a shopping spree, and nominal GDP growth will accelerate. (Arguably, this would likely be in the form of inflation.) The denominator of the debt-to-GDP ratio will rise more rapidly than 5% per year, and the debt-to-GDP ratio will fall.

If that argument sounds like hand-waving, there is a simpler argument. If the debt-to-GDP ratio gets arbitrarily large, at some finite time point, we must have a situation where one individual owns government liabilities that are 10000% of national GDP. That one person could buy up the entire national output for the year by just using 1% of his or her wealth. We then need to ask: is that a plausible outcome, or would prices adjust in such a fashion to make that impossible?

In other words, the assumption that the inflation target would break before the debt-to-GDP ratio gets extremely large.

In order to allow the debt-to-GDP ratio to get large, we need to somehow have very low nominal GDP growth coupled with high fiscal deficits. Japan and some European countries managed to pull off that feat, but even then, net debt-to-GDP ratios are not going to infinity. (Japanese government accounting is baroque, and the Japanese government is itself the largest holder of Japanese government debt. This results in a very large gap between gross and net debt figures for Japan. Since a debt that a government owes to itself has no economic impact, there is nothing stopping it from going to infinity.)

The problem with these partial models is that there is no mechanism to resolve what happens when the debt-to-GDP ratio gets "too large." Inflation (or nominal GDP) is typically assumed to follow a fixed path, and so there is no mechanism for behavioural feedback to correct the debt-to-GDP trajectory. We need to go to a full economic model to resolve the impasse.

Full Models

In order to judge what mechanism will prevent the debt-to-GDP ratio from "going to infinity," we need a full economic model. The inability of MMTers and the mainstream to communicate on this topic largely reflects the modelling tradition.

The mainstream model arguments will generally look as follows.
  • Within the model, the central bank sets interest rates to keep inflation on target. Everyone assumes that the central bank's resolve is credible, and so inflation is forecast to remain at the target level for all time going forward. (There could be shocks that cause temporary divergences, but those divergences are expected to be clamped down by the central bank reaction function.)
  • Since inflation cannot move -- by assumption -- fiscal policy has to be set to allow this. There is the inter-temporal governmental budget constraint which allegedly implies that the inflation target is credible. That is, there is a mathematical test on forecasts for fiscal policy that must hold (or else?). Importantly, there is no trade-off between inflation and fiscal policy -- by assumption.
On the MMT side, the modelling is more eclectic. The tendency is to use existing stock-flow consistent (SFC) models from the post-Keynesian literature, and then modify them to be useful for the task at hand. Although it might be a desired property that the inflation target is hit, there is no reason that it must be hit. A loose fiscal policy will result in higher inflation -- and so there is an actual trade-off between fiscal policy parameters and inflation.

In summary, the mainstream view of sustainability is stuck in a binary sustainable/unsustainable condition, whereas MMTers (and others) view it as a trade-off. This results in largely pointless debates on Twitter. This also explains why mainstream authors always invoke things like "fiscal crises" that trigger on unknown conditions, which is then scoffed at by MMTers.

In order to have some common ground, we need to look at shorter time horizons. Take for example the Republican tax cuts in the United States. It is not hard to find both mainstream economists and MMTers who argue that these tax cuts will result in higher inflation. Meanwhile, economists who are fans of the tax cuts will disagree about the inflation risks. It is possible to have a useful debate on this topic, because even the most dyed-in-the-wool mainstream economist realises that it would be silly to argue that these tax cuts will cause an immediate collapse of economic equilibrium because the transversality condition is not met. In other words, they drop their binary world view, and look at the deficit/inflation trade-off (like the MMTers). Unfortunately for most of the mainstream models, they have assumed fiscal policy out of existence within the models (monetary policy always perfectly offsets it). This makes it somewhat difficult to estimate the effect of fiscal policy on the economy.

However, even if the budget constraint is dropped from mainstream theory (and it is dropped within certain classes of models), there will still be an impasse. The logic of mainstream models is that prices are uniquely fixed by various conditions on the derivatives of variables (modulo the Calvo fairy effect). This limits the path for fiscal policy if we assume that the inflation target is hit at all times. The post-Keynesian tradition argues that price determination is much more uncertain in practice, and so there is more space to vary fiscal policy and still hit the inflation target.

Concluding Remarks

I will return to the mathematics of the governmental budget constraint in future articles. That is taking on the mainstream view on its own mathematical turf. However, if we put that part of the debate aside, we can see how the different theoretical world views results in both sides talking past each other in debates.

(c) Brian Romanchuk 2018


  1. What is the mainstream these days? It wasn't so long ago that it was the New Classical paradigm. More lately, it looks more like New Keynesianism with a GE substrate? If it was the former, then there would be little room for debate given Ricardian Equivalence. If it's the latter, then there is some room for commonality if you can get over the use of budget constraints which assume a given level of income and asset structure. With a given level of income, it seems to me that you can't talk about levels of employment and unemployment.

    Henry Rech

    1. It’s hard to generalise, yes. The fact that it’s a “budget constraint” is what turns it into a binary condition - it either holds or it doesn’t. There’s no notion that a progressively looser fiscal policy causes higher inflation - because of monetary offset. I don’t think anyone has abandoned monetary offset within the “mainstream”.

      The closest to common ground might be the Fiscal Theory of the Price Level. That’s a minority view, and invalidates the belief in monetary policy dominance. However, since the price level did not jump when the Republican tax cut was passed, it takes a certain amount of crazy to take the FTPL seriously.


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.