Primer: Post-Keynesian Inflation Theory Basics

This article is an introduction to the post-Keynesian approach to inflation. It is largely based on Section 8.1.1 of Professor Marc Lavoie's Post-Keynesian Economics: New Foundations (link to my review). Similar to the work on stock-flow consistent models, we start out with what is essentially an accounting identity: a statement that is true by definition. We need to understand the implications of the accounting identity before we worry about the behavioural aspects (which are not pinned down with accounting).

(The approach here is quite distinct from conventional approaches; I discussed why post-Keynesians reject conventional inflation theory in an earlier article.)

Markup Pricing

For simplicity, this article only discusses a closed economy; that is, an economy that is not open to external trade. Section 8.1.2 of Post-Keynesian Economics discusses the open economy case.

In a monetary exchange economy, pretty much all goods and services that are exchanged based on monetary transactions are the results of some form of labour. Yes, capital goods are typically required, but those capital goods were the result of labour inputs. For those of us worried about resource consumption, commodity inputs are important, but those commodities do not magically appear -- human labour is required to extract them. If one wanted to paraphrase some Teutonic economists, goods and services are the result of human action.

We can use Hollywood dystopian fiction to give a good illustration. On Earth, we do not (yet!) pay for access to oxygen; it is available to healthy individuals without any conscious effort. Whereas if we turn to the (presumably) fictional Mars of the Arnold Schwarzenegger version of Total Recall, oxygen was under the control of the rather villainous entity that ruled Mars. (Admittedly, this breaks down at the end of the film, where oxygen was made available by alien technology.)  So oxygen is outside the analysis of economies on Earth, whereas it would be important on (pre-alien intervention) Mars.

In a capitalist society, the bulk of output is by paid labour. Although there are self-employed individuals (like myself) and worker-owned firms, they are still a minority. The defining characteristic of workers is that they are paid a wage, which is normally fixed nominally (although employees at the top of the hierarchy skim off a lot of profits via bonus schemes).

If we assume that all output is the result of wage labour, we can arrive at the identity (due to Weintraub):
\[
p = \kappa \frac{w}{y},
\]
where:

• $\kappa$ is the average markup;
• $w$ is the nominal wage rate;
• $y$ is the output per worker.

For example, if $\kappa$ is 1.1, then the selling price output is 10% higher than the wage cost to produce it. 

Rates of Change

We now turn to the rate of change of prices -- inflation.

We denote the percentage change of a variable $x$ as $\hat{x}$, which is defined as:

\[
\hat{x(t)} = \frac{ \frac{dx}{dt}}{x(t)},
\]
(the rate of change of $x$ divided by $x$ itself). Needless to say, we assume that the variable stays away from zero. (Unfortunately, the hats on my variables seem to be offset when the page is rendered, which is annoying.)

If we apply some basic calculus, we can see that:
\[
\hat{p} = \frac{\frac{dp}{dt}}{p(t)} = \hat{w}(t) - \hat{y}(t) + \hat{\kappa}(t).
\]

Interpretation

We want to be able to explain persistent inflation; for short periods of time, the three terms in the rate of change identity can do all sorts of things.

The argument in Post-Keynesian Economics is that markups cannot rise forever, as that would imply an ever-rising profit share of national income. One might cynically note that this is exactly what has been happening for the post-1980 period, but even so, the rise in the price level was much greater than the rise in profits over that period. Instead, we need to look at the first two terms: how much greater wage growth is than output per worker (${\hat w} - \hat{y}$).

The analysis then leads to: why will wage gains outstrip productivity? The post-Keynesian answer is that this will happen if workers' bargaining position increases relative to that of business owners. As I noted previously, coming up with a mathematical formulation of the notion of "workers' bargaining position" is going to be difficult.

However, one can see the attraction of this theory. By most accounts, the bargaining position of labour has been crippled as a result of structural changes imposed since the early 1980s. From this standpoint, the deceleration of inflation is no accident.

How Different is This?

If we stick to just the accounting identity that I described in this article, one could reasonably argue that this is not that much different than conventional theories about inflation. The same accounting identity appears, since it is obviously true.

The difference is stark if we look at political economy considerations: a viewpoint that has largely been whitewashed from polite economic conversation. If we stick to the mathematical formulation used in mainstream macro, prices are purely the result of marginal productivity; the concept of a struggle for income shares disappears by definition. (One may note that not all mainstream economists entirely buy that story, even if they are otherwise theoretically orthodox.)

From a practical consideration, central bankers are obsessed with labour market statistics. Is the unemployment rate below the dreaded NAIRU? Although the output gap -- as defined by actual GDP versus "potential" -- is allegedly more useful than the "NAIRU gap," one may note that it has barely come up in conversation this cycle (other than being invoked by rate doves). In previous cycles, the output gap was more popular in market analysis. No matter what mainstream economics allegedly says about the drivers of inflation, wage inflation is what matters in practice.

However, one may note the absence of the notion of the central bank determining the price level, or the diminished role of expectations. Since administered prices are set by human beings as a markup over costs (a point I noted in an earlier article), their expectations about their future costs obviously matters. For this reason, we should not be surprised that there is relationship between inflation expectations and realised inflation. However, it is unclear how far we can run with that concept.

If one demands a reduced order inflation model, we need to jump to Section 8.4 of Post-Keynesian Economics -- the conflicting claims model. However, despite my training as an applied mathematician, I am deeply skeptical about such reduced-form models. I have less objections to the post-Keynesian version. The reason is that the models contain some fairly arbitrary variables, such as targets for real wages. Why would workers' target for their real wages change? Although one might be able to do some historical analysis, it seems straightforward that such variables are a function of the political economy environment. We can try fitting a model to historical data, but any extrapolation of past trends is obviously contingent on the environment not changing. This is not a feature of mainstream macro, which allegedly capture universal truths about the economy.

If we return to my complaints about price index aggregation (detailed in my earlier article), we see that the simplified post-Keynesian models are also hit by them. (Within the post-Keynsian literature, there is a lot of detailed empirical work on inflation, which would be compatible with the price level aggregation critique.) Although the aggregate accounting identity is true by definition, it is the result of lumping together very different price measures. A good example is the runaway cost inflation in American university tuition: it is certainly not the result of paying line professors too much. We would need to attempt to apply the concepts to the components of the domestic price structure where the price is best explained by a markup over line employee wages; other parts of the CPI will march to different drummers.

From a policy standpoint, we see why the government needs to be concerned about wage growth in the context of inflation control. Real output per worker is set by the "technology" of the production process; it is unclear how much we can increase that over time (although every political party tends to promise that this is what their programme will deliver). Although the government should be neutral with regards to the distribution of income between labour and capital, it does seem somewhat implausible to worry about an ever-increasing level of profits with a fixed level of wages. By the process of elimination, this means that policy makers need to ensure that wage growth cannot indefinitely outstrip productivity growth (which is assumed to be outside the control of policy). This logic explains why I would only be concerned about a return of a 1970s style inflation in the context of changes in the political environment.

Concluding Remarks

From the perspective of my potential book on the business cycle, I doubt that I will go much further down the rabbit hole of inflation theory than this. From the perspective of orthodox mainstream macro, such a decoupling of inflation from the business cycle is insane: the determination of prices is the whole key to the business cycle. However, the evidence of such a view is sketchy at best, and so I would prefer to move inflation analysis to a later report.

(It may be that I will eventually bind my reports into a single big book, but I would rather keep my reports at a manageable size.)

(c) Brian Romanchuk 2018