The following sentence appears in the abstract.
Critics question the sustainability of MMT-prescribed approaches to fiscal and monetary policy, especially over extended periods of time, in the presence of international financial markets, and for developing country governments that borrow in foreign currency.
This ought to immediately ring alarm bells in readers’ heads. From the perspective of idealised mathematical models, a great deal of MMT debates revolve around “currency sovereigns.” In the real world, actual sovereigns lie on a somewhat messier “currency sovereignty spectrum.” In some cases, one could have a good faith debate whether a particular country is at least close to being a currency sovereign. However, there is not doubt that any country that borrows in foreign currency to any significant extent is not a currency sovereign.
(Why the “significant extent” wording? The Government of Canada is very close to the “currency sovereign” end of the spectrum, but it does have some outstanding foreign currency bonds. These bonds effectively finance Canada’s relatively insignificant foreign currency reserves, and also creates the administrative capacity to borrow in foreign currencies if absolutely necessary. To the best of my memory, this is common practice, with the United States being a major exception.)
If one wants to know what MMT proponents have to say about countries that are not currency sovereigns, you would need to read the papers that address that topic. My personal interest is solely the countries which I followed as a fixed income analyst, which are the developed market sovereigns. These include currency sovereigns as well as the euro peg countries. I am unfamiliar with what be termed the “emerging market” MMT literature — but I cannot see citations of emerging market MMT papers in Razmi’s article either.
What Does the Model Say?
I approached this paper in exactly the same way that I approached published papers in control theory when I was in academia: I jumped straight to the mathematical parts, and largely skipped all the textual representations about what the mathematics means. I am an applied mathematician: I can see the implications of a mathematical myself, I am not interested in what the author says that they mean. In economics particularly, there is often a massive gap between those two positions.
As a result, I am not going to quote the article extensively. The author includes a potted history of the MMT debates, although I am extremely unimpressed with the critiques cited that I am familiar with. Someone new to the debate might wish to use these references to start a literature survey. My interest is in what is new in this article itself, and the new content is based on the mathematical model and its implications.
The immediate problem is that the model is one of which in countries borrow in a foreign currency. This means the model is useless for a currency sovereign. As for non-currency sovereigns, we need to immediately ask: what are the MMT policy prescriptions for such countries? To what extent that could be puzzled out from the Razmi paper, the article’s assumption is that the policy prescriptions are the same as for currency sovereigns. This then raises the question: if MMT proponents believed that the policy prescriptions are the same for currency sovereigns and non-sovereigns, why do the repeatedly state that being a currency sovereign opens policy space?
Since I have not studied that part of the MMT literature, I will not attempt to figure out whether Razmi correctly characterised the MMT literature on non-currency sovereigns. Given that the policy prescriptions that Razmi discusses matches what MMT primers say about currency sovereigns, my hopes are not particularly high.
No Risk Assets
The paper relies on portfolio balance effects in a world where the only assets are local currency bills and foreign currency bills. Although algebraic tractability is going to be an issue for adding more asset classes, this is a very misleading situation. In the real world, currency market “clearing” is generally dependent upon risk asset flows — foreign direct investment, equities, corporate debt, real estate, and ape jpegs. The only time that government security flows are dominant is the case of countries that intervene in currency markets in size. This could be either a peg to hard currency, or a “managed float,” as seen in Japan’s history after the folding of the Bretton Woods system.
Although the case of countries managing their currency versus a “hard currency” is interesting, it goes without saying that the currency in such a case is not in fact floating. It certainly is not the focus of the MMT literature that I am familiar with, other than the questions around the impact on the “hard currency” issuer (the “Is China financing the United States?” debate of the previous decade). For what it is worth, I think that the export-led growth strategy built around a permanently undervalued currency (originally employed by Germany and Japan) is a good development strategy, but it is reliant upon geopolitical conditions (the willingness of the United States to absorb those exports). As such, I am unsure what the value-added of discussing such strategies is in 2022.
From the perspective of both developed countries and developing countries, this model structure is a serious defect.
In developed countries, governments do not finance themselves in truly “foreign” currencies (with the euro peg sort-of exception, which has been extensively looked at in the MMT literature). Even in the private sector, competent lenders do not finance borrowers with a currency mismatch. Post-Keynesians are a consistent source of absolutely terrible analysis of developed countries since they insist that borrowers love issuing mismatched foreign currency debt, even though they are obviously incorrect (where are all the defaults caused by currency movements?).
In order to capture the full range of emerging market crises, we need risk assets to create financial instability (as per Minsky). I am not an emerging market expert, but most of the conflagrations that I can remember resulted from pro-cyclical risk asset flows. There may be some countries that can reliably control their currency value with interest rate policy, but I am somewhat skeptical as to how widespread that is.
Although the belief that interest rate policy can influence the level of a currency, it is a bit of stretch to assume that a country can control its currency value solely by changing interest rates. If that were indeed possible, anyone able to do a half decent job of predicting interest rate differentials would be able to make a mint trading foreign exchange. The reality is that many investors do build “carry portfolios,” but the strategy works only until it doesn’t. (We can also note that the Russian ruble has collapsed despite a doubling of policy rates to around 20% at the time of writing.)
Inflation Pass-Through From Currency Changes Country-Dependent
The toy models used by post-Keynesians in these analyses suggest that there is a large pass-through into domestic inflation from currency changes. However, if we look at the developed countries in the “modern era”, it is extremely difficult to see the effects of exchange rate changes. In a services-based economy, the major driver of firm costs is the wage bill. Energy costs are admittedly set in global markets, but the volatility of energy commodity prices are much larger than the volatility of developed exchange rates: if oil prices spike, they are going up in all the developed economies, even though currency quotes are changing.
Of course, this is not true for all countries. Less developed countries where unprocessed food are a large proportion of consumer expenditures and rely on imported manufactured goods will have more exchange rate pass through.
Although it might not seem intuitive, this factor is best understood as being a component of “currency sovereignty”: can a country pursue a policy of benign neglect of the value of its exchange rate? The developed countries largely have this flexibility, whereas this is not the case for many developing countries. The implication is that developing countries have less policy space, and that has nothing to do with the currency in which the government borrows. Policy recommendations need to reflect this reality. Returning to the article we are discussing, if we want to discuss MMT in this context, we would need to read MMT proponents’ discussion of what to do in this situation — and not discussions that are explicitly or implicitly aimed at developed countries.
MMT Policy Recommendations Ignored
The model in the Razmi paper ignores key MMT policy recommendations. The most important of which is the Job Guarantee, which is completely unmodelled. It makes no sense to argue that MMT policy recommendations are in some sense “simplistic” if you do not make any attempt to model them.
How one would model a Job Guarantee is going to be controversial. However, unless the Job Guarantee wage is set at so low a level that it would be ignored by everybody, I would expect that at least 2% of the population would be employed by the programme, even during boom times, but the average across the cycle might be closer to 4-5%. (I want to emphasise that these figures are purely my guesstimates, and do not reflect any scholarly estimates.)
The Job Guarantee programme would thus be significant enough to effect wages at the low end of the wage scale. As a result, the Job Guarantee wage is a policy variable.
In any event, the modelling tools used within the paper are worthless when it comes to analysing the Job Guarantee, since it ends up being stability analysis around some stationary point. Either the analysis suggests that the Job Guarantee wage has no effect on aggregate wages/prices — which is obviously nonsensical — or else it would suggest that governments can fine-tune the economy just by changing the Job Guarantee wage. (Why is the no effect possibility nonsensical? It implies that the government could raise the Job Guarantee wage to an arbitrarily high level without affecting average wages.) Although MMT proponents are fans of using the Job Guarantee wage as a policy tool, I doubt that any of them think it can be used to make economic variables follow arbitrary target trajectories. Realistically, the effects of the Job Guarantee are going to be highly nonlinear (e.g., much easier to prevent deflation in wages than inflation), and we cannot discuss the macro effects of the programme with a stability analysis of a steady state.
Meanwhile, the simplified aggregate model ignores MMT arguments regarding the use of institutional policy levers in an attempt to control inflation. Such policy levers are complex and controversial (and unpopular with conventional economists), but we cannot truly discuss the MMT position without them. Once again, the modelling technique in the paper cannot be used to model the MMT position.
A mathematical model is just a set of statements about sets. One makes assumptions, then one cranks the handle of mathematical operations to see what is implied.
Many simplistic Post-Keynesian and neoclassical macro models are built around an assumption that government policy is largely captured by two policy levers: interest rate policy, as well aggregate fiscal policy. (“Money creation” might appear, but so long as the money demand function is monotonic in interest rates, a money growth rule is just an alternative expression of an interest rate rule.)
It is therefore unsurprising that if you take such a model, and you then remove interest rate policy, the policy space is greatly reduced. However, the questions that matter are whether that model reflects reality, and/or whether other policy levers exist. There is no point in adding epicycles to models in already published papers if you make no attempt to model what you are supposed to be modelling.