Recent Posts

Tuesday, September 5, 2023

Inflation And The Labour Market

I have been seeing comments about how labour market models have been misleading this cycle. The fun thing about this subject is how mainstream economics is supposed to be a rigorous data-driven science, yet mainstream economists are flailing around trying to come up with a relationship between two time series. (Note that this article is re-hashing points I made previously; I am too lazy to check whether I am just re-writing an old article.)

The usual response to critics who state things like I just did is “give me a better model.” The idea is that we need to replace one reductionist model with another reductionist model. The reasoning seems to be that economics is like physics, where a lot of the history of the field is doing exactly that. (Physicists might be getting into “complexity,” which may or may not be a mathematical pseudo-science. In any event, this is not what people have in mind when they compare economics to physics.) If inflation is a complicated process, any reductionist model is going to fail.

Any empirical work on the link between inflation and the labour market is going to run into a snag that is the result of what I argue are relatively non-controversial positions.

  1. We assume that there exists a unitary variable that summarises the business cycle — which might not be directly measurable. It might be something like the first principal component of a few underlying variables. The usual measured variable that is supposed to be a proxy of this variable is GDP, but one may note that the NBER looks at a variety of variables to date recessions. The justification for the existence of this variable is that recessions are a somewhat nonsensical concept without the notion of some way of summarising the state of the business cycle which goes up and down. (If one wants to insist that no such unitary variable exists, we end up with business cycle nihilism. This might be the correct stance, but makes it very awkward to discuss macro.)

  2. Inflation is a pro-cyclical variable, possibly with a lag. This can be justified by eyeballing inflation/GDP charts. There are any number of theoretical stories justifying why this is the case.

  3. Employment growth is pro-cyclical, almost by definition. Rising unemployment is one of the defining characteristics of a recession.

  4. (Core/median) inflation and employment growth empirically exhibit “trends” — although monthly data might be noisy, the averages across months tend to be smoother after seasonal adjustment (albeit with step changes during things like recession).

  5. As an added bonus, we can note that private sector credit is pro-cyclical. This can be seen by eyeballing charts, and is likely to be a component of any model that takes the Kalecki Profit Equation seriously. (Neoclassical models notably do not.) Bank lending and hence deposits are a component of private credit, and thus the bank deposit component of M1 is going to be pro-cyclical.

You do not need a doctorate in statistics or need to study Real Analysis to know that 1-4 taken together imply that inflation and employment growth are going to be correlated to some possibly unknown “business cycle” variable. It is going to be nearly impossible to detect a causal relationship between the variables unless the relationship is simple and stable over time. Guess what? We cannot find any such relationship.

(The fifth point in the list is aimed at any Monetarists who for some bizarre reason decided to read this article.)

Inflation is complicated, and there is no reason to expect that rents are going to move in the same way as college tuition, used automobiles, or imported jeans. To any extent we can model inflation, we need to decompose the aggregate (which was allegedly proved to be the wrong way to look at inflation, according some neoclassical economists).

The whole “unemployment needs to rise to reduce inflation” was a terrible take since it was a somewhat innumerate understanding of a “stylised fact” about recessions. Recessions tend to be associated with disinflation (rate of inflation dropping). As such, one strategy to control inflation is to throw the economy into recession whenever inflation is “too high.” (I am not saying that is a good strategy, but reading between the lines, that is the neoliberal strategy.) However, the causal implication is one way — a recession is (typically) sufficient for disinflation, but it is not necessary (as seen in the disinflation after the pandemic spike).

Book Progress? Sigh.

I am still plugging away at editing my inflation manuscript (interrupted by the CFL Labour Day Classic). Most of the work is tweaking existing text, and thus not publishable here. However, I added a missing section that should show up within a week or so.

If I were productive, I might be able to finish it off within a month or two. Based on past experience, a publishing date in January is more realistic.

Email subscription: Go to 

(c) Brian Romanchuk 2023

No comments:

Post a Comment

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.