Recent Posts

Tuesday, August 25, 2020

Replicating MMT In A Neoclassical Framework?

One of the standard criticisms raised by many neoclassical economists is that there is nothing new to MMT. Given that neoclassical economists and MMT proponents have been arguing loudly online for over a decade, it seems hard to support the view that MMT is a subset of neoclassical economics. (Another weak point in the argument is that the person making it rarely has read any MMT academic work, so the critic is not exactly sure what features are part of neoclassical economics.)

A slightly less adversarial stance taken is that MMT insights can be replicated by adapting neoclassical methods. I think the answer to that question is: yes and no. Although it seems possible to get parallel results to parts of “narrow MMT,” but the integration with broad MMT does not appear to be possible. This is more plausible, and the subject of discussion here.

(Note: This is an unedited draft section of my MMT primer. It is part of a chapter discussing common critiques of MMT.)

Policy Conclusions: Yes, They Can Overlap

In online discussions of policies, it is quite often possible that a neoclassical economist and a MMT proponent will have the same policy views. In particular, at least among the center-left, there is an acceptance that loose fiscal policy was the best response to the collapse in activity in 2020. Furthermore, the previous neoclassical consensus that monetary policy alone was sufficient for economic stabilisation has weakened.

Having policy views overlap is not particularly surprising. Only free market ideologues now disagree with the view that loosening fiscal policy will increase nominal GDP. As such, one can arrive at the same fiscal policy stance from any number of theoretical directions. If one is concerned with political coalition building, one probably should emphasise the areas of common ground.

However, even if economists agree about a basic principle about policy, there is no guarantee that they will agree about implementation details. In the pandemic crisis, the lockdown conditions greatly limited organisational capacities, and there were extreme time constraints. Under those conditions, each country had very few reasonable options to choose from, and the choices were determined by pre-existing institutions. As such, the scope of disagreement between economists that occupied similar points on the political spectrum was limited. As policy freedom returns, the room for disagreements will grow.

In any event, if one is interested in theory, one needs to be a stickler for theoretical details, and mushing together logically incoherent systems is a bad idea.

Narrow MMT

Many neoclassical economists focus on the policy recommendation of eliminating government bond issuance (Section 3.5). This policy can be captured in a neoclassical model, and from that extremely narrow perspective, they believe that they have captured the essence of MMT. The problem is that this is only one policy recommendation and does not even cover all narrow MMT (never mind broad MMT).

Instead, the closest a single model could do is to replicate the Monetary Monopoly model in a neoclassical framework. (This may have already been done.) Given the simplicity of the model, one could start from the description in Mosler’s White Paper (Section 4.2) and build it.
There are issues when adapting neoclassical micro-foundations to the Monetary Monopoly Model. In the model, a key price is set by government fiat, and thus cannot be thought of as the result of an equilibrium process.

If the government is pegging the price of a good, things are trivial if we assume that is an aggregate good and the “law of one price” holds. The government sets the price level in every period and the price level can follow any arbitrary trajectory chosen by the government. This would result in an extremely silly model, and MMT proponents like myself would mock its reductionist nature. To be a semi-plausible, more than one good would need to be introduced.

However, pegging the price of a good is not the usual MMT policy prescription, rather it is the Job Guarantee (where the JG wage is pegged).

In this case, the usual micro-foundations of the household sector would need to be adapted. The usual framework is that households decide upon the number of hours worked, trading off leisure time versus the consumption possibilities offered by increased total wages by working longer hours. (The implication is that recessions are just workers voluntarily taking vacations, which people like me mock whenever possible, such as I am doing right now.) The Job Guarantee is not supposed to be a vacation; it is work. As such, it is meant to be fungible with working in the private sector. The problem is that unless the private sector wage is assumed to be equal to the JG wage, the optimal solution is to not work in the Job Guarantee. (If the private sector wage is assumed to be equal to the JG wage – the law of One Price – then the macro model ends up being almost as trivial as the one in which the government sets the price of the composite good.) This does not align with what seems to be the reasonable expectation: if the JG wage is set at a living wage, unemployed people will show up to work in it.

The end result if that a kludge has to introduced to the micro-foundations: there would have to a purely arbitrary function that splits the hours worked between the Job Guarantee and the private sector, which would presumably be a function of capacity utilisation and the wage differential. This type of kludge is the sort of thing that neoclassical economists insisted was unacceptable in economic theory publications when they seized the means of academic production.

Once this kludge is introduced, it seems entirely possible that the resulting model would end up looking similar to an aggregated model produced by a post-Keynesian (with theoretical differences that only post-Keynesians care about). It seems safe to say that if a Job Guarantee was a serious political possibility in one or more developed countries, such a model might appear.

Assuming such a model was developed, it is a safe bet that it would be the focus of interest for mainstream economists. Since it does not yet exist, I obviously cannot say how accurately it represents MMT thinking. As the next section will elaborate, it will certainly not represent all of MMT. However, for something like the analysis of a Job Guarantee, having more bodies around the table doing empirical work has to be a welcome development.

Broad MMT: Not Really

Realistically speaking, forcing a Job Guarantee into a neoclassical model is as far as the convergence would go. Beyond that, we are drifting into Broad MMT. As discussed in Section 4.5, the only thing that post-Keynesians can reliably agree on is that practically every aspect of the neoclassical theoretical tradition is incorrect.

Neoclassicals do relax some of their questionable theoretical assumptions, but the framework only accepts a small number of changed assumptions at a time (as tractability of the model will break). Throwing every assumption into the trash bin simultaneously is a step too far. Furthermore if every single contested assumption is eliminated, it is very hard to define the resulting model as “neoclassical” – it would just be a post-Keynesian model that has been re-written.

That said, one could imagine that a theoretical revolution on the order of the transition from Old Keynesian thinking to New Keynesian thinking could happen, and the neoclassical foundations are warped in a fashion to incorporate post-Keynesian criticism.


Based on the tendency of neoclassical authors to produce floods of papers on any topic, it seems entirely possible that the neoclassical framework will be adapted to generate results that appear similar to MMT – at least from the perspective of the people who built the models. If we are lucky, such developments could allow for more reasonable debates.

However, in the absence of such work, there are enough theoretical divergences that imply that existing neoclassical theory does not cover all insights from MMT.

(c) Brian Romanchuk 2020


  1. You can take an agent based model purporting to replicate mainstream macro and add the Job Guarantee to it.

    Work may be underway on that...

    Except that mainstream macro doesn't appear to be that robust when you start introducing reality.

    1. Interesting, but I guess that would be viewed as changing the subject. You gotta have a model with agents that forecast the future, yadda, yadda, yadda...


  2. Very interesting article. Thanks for sharing!


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.