A MMT View On The Theory Of Hyperinflations

The post "CMMT - Cate's Modern Monetary Theory" (on Seeking Alpha; free registration required) by Vincent Cate attracted a fair number of comments on the Mike Norman Economics web site. In that post, he presents what he calls CMMT - Corrected Modern Monetary Theory - and he argues that this correction allows MMT to explain hyperinflation. One could probably argue that MMT - as well as mainstream economic theory - do not have standard models that deal with hyperinflation. But that is for the same reason that those bodies of thought do not have standard models to estimate the impact of barbarian incursions along the frontier. It is not to say that foreign incursions do not matter - as the Western Roman Empire can attest - but that such an event is not a serious concern for the industrial economies at present.


I will not attempt to deal with the mechanics of Vincent Cate's model. It appears to based on the quantity theory of money, and it is easy for the reader to validate that the quantity theory has little empirical support. Instead, I want to discuss the theory of hyperinflations.

Hyperinflations - Not Just High Inflation

A hyperinflation is not just a high inflation; it is a situation where the price level rises by at least 50% per month. This Wikipedia article offers a good overview, and lists the historical cases. If you look closely, there are no developed countries with free-floating currencies in the list of hyperinflations. This is no accident; there are strong institutional factors in developed countries that prevent hyperinflation. But if you read articles on some corners of the internet or read market commentary, hyperinflation is always on the verge of breaking out.

The extremely high rate of inflation is an extremely important factor in distinguishing a hyperinflation from a mere "high inflation" episode (which many developed countries had in the 1970s). I discuss that point further below.

Is Hyperinflation In A Developed Country Possible?

Various MMT economists have written about hyperinflations, and they have tended to correctly point out that hyperinflation is associated with a country with large foreign currency liabilities and a collapse in its productive capacity. In other words, a hyperinflation is just the death throes of a economy. (Note that although I have MMT leanings, my views are not necessarily 100% aligned with MMT. As such, I am not a "spokesman for MMT".)

I have not seen any MMT economists pursue the line of thought too far, but it would be possible for a developed economy to have a hyperinflation in the absence of a shock which wipes out its productive capacity. However, it would take a huge amount of stupidity as well as the ability to drive economic policies through safeguards that are put into place within representative governments.

For example, assume that I was invited to induce a rapid hyperinflation in Canada for some reason or another, and I was given the constitutional authority to ram through policy changes for all levels of government. But to make it slightly harder, we'll assume that I cannot be too blatant about it (I cannot do something obvious like mail everyone cheques for $1 billion Canadian dollars) and that the Bank of Canada keeps its 2% inflation mandate (so I cannot just wave the inflation expectations magic wand). I would:

• Abolish all sales taxes and the value-added taxes (called the GST in Canada). They would be replaced by taxes with fixed dollar amounts (such as a 50 cents a litre tax on gasoline).
• Income tax withholding would be abolished; income taxes would be paid by the middle of the following calendar year.
• All government payments would be indexed to current official U.S. dollar exchange rate, and the quotes used would be updated as quickly as possible. (If the Canadian dollar somehow strengthened, the indexation factor will not be allowed to drop below 1, to prevent a "hyperdeflation".) Alternatively, payments could be indexed to a commodity, like gold. We cannot use the CPI, as it is not calculated fast enough to get a good hyperinflation going.
• To get the ball rolling, I would raise unemployment insurance and welfare payments by 10%. 

The end result is that government revenue would be based on essentially fixed taxes, while spending would indexed to the U.S. dollar. The Canadian dollar would drop in response to a larger fiscal deficit, and the indexation would cause those deficits to blow out further. It would end up worthless in a matter of months. Meanwhile, businesses would switch over to using the U.S. dollar as a unit of account. (They would not use gold, as they would need access to U.S.-denominated loans; there is no market for gold-denominated loans.)

[UPDATE: I have made some changes to the above paragraphs to clarify points in response to comments.]

The central bank and the rentier class could attempt to induce a recession to prevent the hyperinflation by raising interest rates. Unfortunately for them, that would just throw more people onto the welfare rolls and increase the amount of interest income for the non-government sector, and the fiscal juggernaut would still roll over them at a later date.

Although the exact steps used could differ, any potential hyperinflation (that did not rely on some external disaster) would probably have to develop in an environment like I described above. Very simply, the automatic stabilisers embedded in the taxation and spending system would have to be deliberately smashed. (Or more accurately, the part of the automatic stabilisers that slow down nominal GDP growth).

A Mainstream View Of Hyperinflation

Within mainstream economic theory, hyperinflations are hard to simulate using workhorse macro models. There are some more specialised models, but those are special cases.

Using the Fiscal Theory Theory of the Price Level as an example, the set of policies I outlined above would imply an "unsustainable" fiscal policy, and the price level would become unbounded ("infinite"). This would happen immediately, and so there would be no hyperinflation to simulate.

Not all DSGE modellers agree with the Fiscal Theory of the Price Level; within those approaches, all that could be said is that the inter-temporal governmental budget constraint (IGBC) was not respected, and there is no solution. There is some loose discussion of "explosive trajectories", which is exactly what a hyperinflation is. My understanding is that such a trajectory is not supposed to happen, so there is not too much of a literature on simulating them.

Although the mainstream approaches do offer a diagnosis for when a hyperinflation can occur (the IGBC is not respected), the condition is too restrictive. For example, take an economy growing at 5% per year indefinitely, with a steady debt-to-GDP ratio and a discount rate of 1%. It does not respect the IGBC, yet there is no reason for a hyperinflation to occur.

A MMT-SFC View

I will now circle back to what Modern Monetary Theory (or a Stock-Flow Consistent Model) says. Modern Monetary Theory is partially a mode of analysis, and it also represents advocacy of particular set of policies. The further a set of policies departs from what MMT advocates, the more the results drift from the way a "MMT-compliant" economy would act. You need to take into account how those divergence will affect the economy.

For example, MMT advocates a free-floating currency. You can use some of the insights of MMT to analyse an economy within the Gold Standard, but you need to understand what constraints are added by the Gold Standard. (Note that MMT is built upon a larger body of Post-Keynesian economics, and so "the insights of MMT" I discuss above may have originated from that larger school of thought. Since I am not an economic historian, I am skipping over these distinctions.)

The set of policies I outlined above to induce a hyperinflation is not an area of interest to MMT economists. They amount to deactivating the automatic stabilisers, which is a very bad idea. Since there is no faction calling for such a set of policies in the developed economies, nobody spends any time worrying about simulating their impact.

But if you wanted to build a model of a hyperinflation, the core idea is that you need to have economic agents set prices using a numeraire* that is not the local currency. In my case, agents would set prices in U.S. dollar terms, but the flows are in Canadian dollars. You then need a mechanism to determine the CAD-USD exchange rate. Since exchange rates are set in markets, they are not easily modelled, but you could probably get reasonable results. All then you need to do is drive the price of $1 CAD to USD $0. The price level (in CAD terms) correspondingly goes to infinity. (I am not familiar with the literature, I believe that this is how the special case of hyperinflation is often analysed. The modelling assumptions will presumably differ between the Post-Keynesian and mainstream approaches. The key difference is that in Post-Keynesian models, prices are administered to be at a level above input costs, whereas in mainstream models prices are driven by marginal costs. A mainstream model pricing is less sensitive to the choice of numeraire.)

This mechanism has to be used, as a hyperinflation is not just a high inflation. The reason is that extremely high rates of inflation cannot be dealt with in the same way as what we are used to.

It must be kept in mind that businesses were able to function for some time, despite the hyperinflationary conditions. Businesses cannot survive if their input costs rise too far relative to output prices, and so there is a limitation of divergences amongst relative prices. As a rough estimate, a "sustained" 1% per month would be a practical upper limit (for a few years). In our current environment, overall inflation rises at about 0.2% per month. This means that individual CPI components can diverge widely relative to the aggregate inflation rate.

In a hyperinflation with 50% monthly aggregate inflation, businesses would still have to keep relative prices close to the aggregate. This means that individual prices have to be fairly closely clustered around the very high average inflation rate. Such a level of coordination would be impossible in the absence of an external store of value. What happens is that businesses are forced to use foreign currency prices, which are then translated into the local currency. (Working from memory, the hyperinflation in Germany ceased during the periods when the foreign exchange markets were closed.)

In fact, this is how hyperinflations are measured. It would have been impossible to calculate a CPI under such conditions; instead academics use the value on the foreign exchange market to back out the implied inflation rate.

Therefore, we see that hyperinflations are actually relatively easy to understand: they are a case when transactions and prices become denominated in a foreign currency, and the local currency is at risk of collapse in the foreign exchange market. Keep out of that trap, and hyperinflation will be avoided.

Footnote:

* A numeraire is the unit of account that is embedded in a model. That is, all prices within the model are expressed as a ratio to the commodity that is the numeraire. A typical choice of numeraire is the local currency, but it can be sometimes easier to chose something else.

(c) Brian Romanchuk 2014