Recent Posts

Monday, January 4, 2021

Transfers And Overheating

Figure: Core Inflation, U.S. and Canada

A recent controversy that erupted was the question as to whether transfer payments (such as the $2000 transfer that was debated in the United States) would cause the economy to overheat. There is an interesting issue: even if $2000 is not enough, what about larger amounts? I have severe doubts that anyone could give a useful answer to what appears to be a simple quantitative question, without a country pushing the envelope for the sake of an economic experiment.

Larry Summers raised this question in a recent op-ed that was dunked on by pretty much my entire Twitter timeline. When I read his piece, I have sympathies with the argument that preventing a collapse in unemployment payments would be a higher priority than a fresh transfer payment. That said, there is nothing stopping pursuing both policies. However, I do not have enough information to say what amount would be a sensible amount to pay, so I am not in a position to pass judgement on Summers positions versus his critics.

Nevertheless, it is interesting that mainstream economists cannot easily answer the question of what level of transfer payments would cause overheating. The ability to answer such quantitative questions is exactly why highly mathematical economic theory was pursued, and more qualitative approaches pushed into the dustbin of editors at mainline journals. To be clear, I am not advancing an inflation model that can answer that question, so my concern is not that they cannot give an answer. Rather, the issue is that an inability to provide a quantitative answer is exactly what heterodox critics had been saying for a long time, only to be met by the response "where is your model?"

Getting an Answer?

I would argue that there are two main approaches to getting a quantitative answer to this question.
  1. Use a model fitted against historical data, and extrapolate the policy change.
  2. Back of the envelope modelling based on a simplified theoretical model.
The issue with the first approach -- fitted models -- is that the fitting has happened in a completely different institutional environment. One could try, but my guess is that it would be a garbage-in, garbage-out exercise. (The justification for that assertion follows from my back-of-the-envelope discussion, which is the remainder of the article.)

The issue with back-of-the-envelope approaches is that each simple theoretical model will give different results. Since it is hard to calibrate against historical data, there is no way to test the candidates in advance. We need to use historical episodes -- and we are getting some data points based on the events of 2020.

The Problem of Inflation

Most empirical analysis of fiscal policy revolves around the issue of multipliers. Different models suggest different multipliers, based on assumed theoretical characteristics of the economy. (In fact, this is what I would suggest.)

Unfortunately, there are two issues with this approach.
  1. The multipliers under current circumstances are likely to be different, given the nature of the current environment.
  2. Even if we get the multiplier correct, that might tell us about nominal GDP over the next few quarters. The relationship between nominal GDP and inflation is not set in stone.

Multipliers are Different

The observation that multipliers would be different in the current environment is well known. With activities closed, even if people want to spend on certain goods or services, they cannot. Meanwhile, businesses that are closed by health edicts are not going to hire, no matter the situation of aggregate demand.

Although medical news has deteriorated markedly in the past month in many areas (including my home province), one has sympathies with a story of a rapid demand increase once vaccine rollouts hit (along with the warmer weather in Northern Hemisphere which coincided with school closures and lower infection rates). So, "the" multiplier might change rapidly.

Link Between Nominal GDP and Inflation Vague

Simple models that rely on a single good (or aggregated continuum of goods) naturally lead to a tight link between production levels and production constraints. In the real world, that linkage breaks down. For example, digital services often have a negligible cost of increased production. (At most, new servers might be needed.) 

This leads to the divergence between post-Keynesian and neoclassical inflation theories. Neoclassical models rely on supply and demand curves, along with some arbitrary price stickiness. Post-Keynesians emphasise that most prices are administered.

Both approaches can come up with an explanation as to why the previous rounds of stimulus have had no inflationary impact (as demonstrated by the chart of U.S. and Canadian core inflation at the top of this article) -- despite the household support programmes involving REALLY BIG DOLLAR AMOUNTS.
  1. Neoclassicals can argue that inflation expectations are anchored because of central bank credibility (or whatever), and so inflation is stable.
  2. Anyone who argues that prices are administered would realise that it makes no sense to raise prices and wages in the middle of the current economic uncertainty. Only sectors with strong supply chain disruptions -- notably, construction -- might entertain it. However, although prices have risen, the construction industry is also using time rationing (delaying projects).
Given that these stories are fairly similar, I see no obvious way to distinguish them. 

Returning to the original question, we need to ask: why would another round of stimulus be any different? So long as the payment is roughly comparable to average households' monthly mortgage or rent payment, it is likely to be absorbed there. Since it is hard to see an immediate hiring binge this far away from vaccine rollouts, second round effects are likely to be muted.

My view is that we could only see more significant effects if the payments are converted to be permanent flows. At which point, those flows need to be compared to regular employment income, and see how significant they are.

Concluding Remarks

One can come up with a reasons to not worry about the inflationary impact of one-time payments, particularly when it is hard to spend on many items. There are obviously limits, but the limits may be far higher than what squeamish politicians and policymakers can stomach.

The story is different for permanent flows. It seems reasonable to believe that the U.S. Federal Government could pay all adults $100/month and the inflationary impact would be close to diddly squat (to use the technical econometrics term). This is less clear once we hit the $1000/month threshold, since that is an income that is a significant portion of many household incomes. My argument is that you would need an actual programme to calibrate numbers closer than that, but the key point is that the metric that matters is the comparison to typical income levels.

Appendix: Rant about Structural Forces

One of the more amusing parts of Larry Summers' op-ed is that he flipped from a secular stagnation story to one where inflation could take off like a rocket as a result of a one-time payment. Not exactly a secular force if it can be defeated by a single policy step.

This is the problem with any "secular" inflation story, such as the ubiquitous demographics arguments (that are just dual y-axis charts with average ages and inflation). There are many structural factors that coincided with the inflation round-trip from the 1970s-2020s in most developed economies. However, it is a hard sell that demographics (or whatever) can prevent inflation if fiscal policy ramps up.

(c) Brian Romanchuk 2021


  1. One model for growing a home garden is: plant seeds and things will grow. It's hard to predict what you might get using this model.

    Another model might focus on growing cucumbers and actually have historical comparisons. The difference between the two models is the number of variables being considered.

    I recognize that you are trying to limit your variables narrowly to "core inflation" but it seems to me that this is like looking at just one row in your garden. Macro economics is composed of many rows of many diverse cultivars.

    My guess is that inflation will 'take off' when government decides to pay people to produce things that people do not want. Those few who continue to produce things people do want can charge what the market will bear, especially if each of the few has some version of monopoly power.

  2. "Larry Summers raised this question in a recent op-ed that was dunked on by pretty much my entire Twitter timeline. When I read his piece, I have sympathies with the argument that preventing a collapse in unemployment payments would be a higher priority than a fresh transfer payment. That said, there is nothing stopping pursuing both policies."

    Unfortunately, at the time Summers panned the increased checks there was something that precluded pursuing the other policy. That would be politics. At that time there was a slim chance to increase those checks- but there was no opportunity for anything else. Those were the choices- bigger checks or nothing.

    Now maybe if the Democrats pull off an unlikely sweep of the Senate runoff elections today in Georgia, then there could be an additional opportunity to do better targeted economic relief. But I doubt they will win both of them.

    1. It looks like the Democratic candidates for Senate did both win. So there is now going to be a chance for Larry to advocate realistically for his preferred more targeted relief interventions. I hope he does- but not holding my breath waiting for it.

  3. I'm still thinking about the issues you raise: Transfers and Overheating.

    What are we transferring? At first glance, only money. But more realistically, we are transferring "wealth". Wealth is always measured in "money".

    So, when government gives away "wealth" equally, everyone's wealth increases (we will ignore who 'really' is giving). It follows that we can divide everyone into two camps: previously wealthy and newly wealthy. It further follows that the newly wealthy camp is more likely to spend new wealth (maybe creating inflation) than the previously wealthy camp.

    Shifting our focus to models, I guess we could assign a model multiplier to this suggested "wealth" effect. We would need additional multipliers for effects from offshore supply sources, Covid-19 restrictions, spending habits shifted by fear, and more, if we wanted to relate the present economic situation to past economic history.

    Not an easy task for any economic model.


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.