Primer: MMT Theory Of Inflation

The Monetary Monopoly Model (Section 4.3) is a core model of Modern Monetary Theory. If we take it literally, we do not need a theory of inflation, since the price level in some variants of the model is explicitly set as a policy variable. However, reality is far more complex than the model suggests, and if we want to match theory to reality, we are faced with the realisation that governments act in a fashion that sabotages the control of the price level that they do have.

(Note: This is an unedited draft of a section of my increasingly long MMT primer.)

The emphasis of this section is the discussion of theory that describes the current environment, with a short discussion of different policy choices.

The post-Keynesian view of the inflation process is that it is complex. There are no simple models that do a good job of fitting reality. This should be obvious to market practitioners – if a simple model did work, it would almost certainly have been found by people doing desperate searches across the set of time series inputs as well as the mathematical techniques to combine those inputs. One side effect of this complexity is that it does not appear that all MMTers will agree on the specifics, and so there is no simple “party line” to describe.

However, this complexity is missed completely by critics, who often invent their own theory of inflation that is attributed to MMT. The most ridiculous version of this I saw was an online assertion that “MMT is the Quantity Theory of Money, with bonds replacing money.” This obviously does not work, and unsurprisingly, there are no references to back the assertion.

Brief Discussion of Post-Keynesian Inflation Theory

The complexity of the inflation process has the implication that not everyone agrees on the details. Those details lie in what I refer to as broad MMT. Section 4.6 discusses broad MMT, as well as split among the various schools of thought within post-Keynesian economics. One would need to delve into these splits to get a better grasp of the variant inflation theories. My explanation here is general, with the focus on what differentiates post-Keynesian thinking from neoclassical.

The first observation to make is that there is a distinction in post-Keynesian economics between flexprice and administered prices, with flexprice being markets that somewhat resemble the simplified supply and demand stories of economics 101.

• A flexprice market is one that is liquid and transparent, with many entities transacting, and prices change rapidly. The clearest expression of these are financial markets, including auxiliary markets like wholesale commodity trading. Commodity trading is often done off exchanges, but exchange pricing often is used as a benchmark. Consumer prices are typically not flexprice, but some qualify – gasoline prices being the most notable, but fresh fruits, vegetable, and meat being other contenders. Online markets that are not acting like traditional retailers also feature flexible prices. These markets are closest to matching the “supply and demand curve” story, but one needs to accept that market pricing is often based on highly variable sentiment (e.g., the spike then crash of oil prices in 2008). The wholesale futures markets are financial markets, and have the odd behaviour that we have come to expect. However, the underlying fair value metric is normally thought of in terms of supply and demand in a broad sense. For example, crude oil futures (and some spot prices) in North dropped to negative prices during the economic shutdown in 2020 once traders realised that there was no available storage to take delivery of oil contract.
• Administered pricing is where the seller sets the price unilaterally, and then fills orders. Prices are normally set as a markup over costs, although there are many strategic choices to make (such as selling “loss leaders” at a low price to drive traffic to a store). The normal preference is to keep these administered prices stable, although some form of indexation can become common at higher inflation rates. The key point is the “supply and demand” curves or “equilibrium” do not appear in the description of the price-setting process.

Since real-world prices are set in a mixture of flexprice and administered markets, aggregate price level determination is complex. For example, energy prices are key input to industrial prices, and can jump around based on sentiment about global growth (or cartel behaviour, as in the 1970s). Meanwhile, things like university tuition are set by administrators, and those administrators can make decisions that appear quite arbitrary.

As based on a response by Nathan Tankus to an initial draft of this text, at least some MMT proponents vigorously dislike the flexprice/administrative terminology. However, since the canonical MMT textbook (as discussed next) distinguishes between domestic prices (presumably administered) versus imported raw materials prices (with international commodities and the exchange obviously flexible prices), there is a distinction between those two types of prices in the literature.

The next principle of post-Keynesian thinking is that inflation is a conflictual process, as discussed in Section 17.3 of Mitchell, Wray, and Watt’s Macroeconomics. Those authors point to the line of authors going back to Michal Kalecki (a Marxist), who see inflation as the result of the struggle over real income. (To understand this, take a simplified model of the economy where everything is static, and all output is sold. The ratio of average wages to average output prices determines the share of output that is captured by workers; the remainder is consumed by capitalists that receive dividends from the profits. Bargaining between workers and firms set that ratio, and both sides wish to increase their share.)

The bargaining position of workers is generally better when the unemployment rate is low, as it is easier to quit. This obviously ends up similar to conventional stories about demand-driven inflation. The differentiating factor is that post-Keynesians view this as a power struggle, and not the result of prices being driven to marginal contributions of the factors of production. The neutering of labour power in the 1990s (Section 2.2) helps explain why inflation has been mild.

Cost-push inflation is another factor highlighted by Michell, Wray, and Watts, with raw material prices being particularly important. They write (page 258):

If in response to the fall in profit margin (mark-ups), domestic firms pass on the raw material cost increases in the form of higher prices, then workers would endure a cut to their real wages.
If the workers resist this erosion of their real wages and push for higher nominal wage growth, then firms can either accept the squeeze on their profit margins or resist.

They summarise the discussion of cost push and demand pull inflation as follows (page 261):

Cost push inflation requires certain aggregate demand conditions for it to be sustained. In this regard, it is hard to differentiate between an inflationary process which was initiated from supply side pressures from one that was initiated by demand side pressures.
[Example of imported raw material price shock omitted.]
However, that conflict needs ‘oxygen’ in the form of ongoing economic activity in sectors where the spiral is robust. In that sense, the conditions that will lead to an accelerating inflation – high levels of economic activity – can also sustain an inflationary spiral emanating from the demand side.

The use of the word “sectors” in the last paragraph is important. It is entirely possible that some sectors are overheating – generating a wage-price spiral in those sectors – even while other sectors are weak and laying off workers. The widespread use of aggregate measures – like the overall unemployment rate – gives a misleading perspective on inflationary pressures in an economy. For this reason, simplified models are expected to fail when applied to real-world data.

From this perspective, one can see immediate problems with conventional price formation stories. Invoking supply and demand is largely an accounting identity or tautology – for every buyer, there is a seller. One needs to be able to come up with some mechanism for predicting price changes, and stories about marginal contributions do not fit real world experience. The reality that most prices are administered also explains why inflation and inflation expectations are linked – if firm managers expect rising input prices, they will raise their output prices. Therefore, the real question is what determines those expectations. Do central bankers have the ability to unilaterally determine an arbitrary expected path of inflation – as suggested by some fundamentalist neoclassicals? Or are inflation expectations based on a holistic assessment of all government policies, as well as realised inflation?

More Distinctive MMT Inflation View?

If we want to draw a stronger distinction between MMT and other schools of thought within post-Keynesian economics, we need to look at discussions of alternative policies.

• The Job Guarantee offers the prospect of an anchor for nominal wages.
• Since inflation is often the result of bottlenecks, one widely-circulated article by Scott Fullwiler, Rohan Grey, and Nathan Tankus suggests using administrative means to deal with rising prices. (The difficulty in describing this approach is that institutional differences are large between countries. The medical and educational systems in the United States have many inefficiencies that can be reduced, while other countries do not have such an abundance of low-hanging fruit.)
• The analysis of the failure of the NAIRU (Section 2.3) tells us what will not work – pretty well any model that uses a single aggregated unemployment rate as an explanatory variable for inflation. (Since employment is pro-cyclical by definition, and inflation under current institutional arrangements is also pro-cyclical, there will be a visual correlation on charts. However, this weak relationship does not offer much predictive power.)
• The microeconomic theory of Fred Lee discusses the logic behind administered pricing. The microeconomic aspect of MMT is discussed further in the section on Broad MMT.

However, if we want to discuss observed inflation, we have to accept that policymakers were not following policies suggested by MMT proponents. The modelling of the resulting inflation is complex, and from a qualitative standpoint, it is often hard to differentiate various views. The rest of this section discusses more distinctive theories associated with MMT, and the difficulty in applying them to explain historical inflation.

The Monetary Monopoly Model

The Monetary Monopoly model does offer a simple explanation of how at least one price is set – it is determined as a policy choice. However, current government policy has been structured so that the government is a price taker in as many areas as possible (as suggested by conventional theory). However, the government does administer some prices, notably the wages of government employees (although public sector unions help determine the wage settlement). One could argue that government policies did accentuate the inflationary period that ended in the 1970s, so the model does fit reality.

My view is that just tweaking how the government requisitions goods and services alone would offer much control over the inflationary process within the current institutional framework. (One can imagine situations where the government might have such control.) Prices are quite often fixed (for periods of time), and governments do attempt to use their bargaining power to get cheaper prices. Nevertheless, cyclical inflation still exists. Instead, by requisitioning labour – the Job Guarantee – the government has much greater policy leverage. Although the previous is my view, one may note that the MMT literature generally discusses the Job Guarantee – or else regulatory/administrative changes (which do not really fit the Monetary Monopoly model story).

Functional Finance: A Story of Limits

Another source of confusion is the discussion of Functional Finance. The pertinent argument from Functional Finance is straightforward: the constraint on fiscal policy is inflation. However, this does mean that we can explain every wiggle of inflation by looking at fiscal variables. The following principles explain why this is so.

• Functional Finance argues that real resource constraints are the source of inflationary pressures, not nominal values. We cannot look at nominal deficit or debt levels and conclude anything about the fiscal stance. We would need to answer the more difficult question of what is happening with real resources.
• If the government is not forcing real constraints to be hit, we cannot say much about inflation based solely on government policy. The private sector still exists, and it can create inflationary pressures on its own.
• An alternative phrasing is that the process is nonlinear. For simplicity, assume that the economy is growing in a steady pace with a fiscal deficit of about 1% of GDP. Let us assume that a tax cut that increases the deficit by 10% of GDP (to 11%) would generate a surge in inflation. This does not imply that a tax cut that generates an increase in the deficit by 1% of GDP would imply an increase of one-tenth the size: if no resource constraints are hit, the modest increase in the deficit will be swamped by whatever else is going on in the economy. This nonlinearity means that we just cannot run regressions with fiscal variables versus inflation and expect them to be useful. Governments have not run fiscal policies that pushed real resource utilisation to extremes in decades, and so one should expect historical correlations to be uninformative.

The argument from MMT proponents is that budgeting frameworks need to be overhauled to account for real resource bottlenecks. This was done in World War II. This task is not trivial, and will not just be a question of calculating a sensitivity between the inflation rate and the fiscal deficit.

No Easy Answers

As I discussed in Section 3.3 of Breakeven Inflation Analysis, forecasting inflation using a mathematical model is inherently challenging. The best results are short-term forecasts that are generated by finding the best model for each major component of the CPI, coupled with guesses about the direction of prices in energy markets (and similar). That said, such models are relatively useful for short-term breakeven inflation trading, but tell us little about inflation over longer horizons.
From a financial market practitioner’s perspective, this should not be a surprise. Inflation is now a traded instrument, and if extremely accurate inflation models existed, the market would be rather boring. Whenever one is confronted with a reductionist model that allegedly predicts inflation, the first question to ask – do inflation-linked traders use it? (The answer is almost invariably: no, they do not.)

As such, the MMT/post-Keynesian story that inflation is complex fits the facts on the ground rather well. A theory that offers a simple model of the inflation process is either tautological (just offers an alternative explanation of something that already happened) or wrong.

References and Further Reading

• Macroeconomics, by William Mitchell, L. Randall Wray, and Martin Watts. Red Globe Press, 2019. ISBN: 978-1-137-61066-9.
• “An MMT response on what causes inflation,” by Scott Fullwiler; Rohan Grey, and Nathan Tankus. FT Alphaville, March 1, 2019. URL: https://ftalphaville.ft.com/2019/03/01/1551434402000/An-MMT-response-on-what-causes-inflation/

(c) Brian Romanchuk 2020