A recent episode of the Debunking Economics podcast annoyed quite a few people, and I couldn’t be happier.
The episode, shown and linked in the graphic below, considered the sustainability of the economic growth that capitalism has spawned over the last 250 years (hint: it isn’t), and as part of it, I took a swipe at Neoclassical economics. A listener tweeted the following excerpt, where I assert that Neoclassical economists don’t have a decent theory of technological change:
And that’s when the fun began on Twitter, with Pontus Rendahl of Cambridge University UK and Gauti Eggertsson of Brown University, USA both weighing in on my lamentable knowledge of economic literature:
I had a dig at Eggertsson for his zero knowledge of the literature on endogenous money (Eggertsson and Krugman 2012) but that’s beside the point here, which is why do economists—including me—frequently not read large slabs of the economic literature?
Eggertsson and Rendahl both imply that, in my case, it’s sheer lack of knowledge. But the reality is much more interesting and revealing about why economics is not a science. It is more like a band of competing religions, where each religion reads its own texts obsessively, but has no idea what the texts of other religions are. So Christians read the Bible, Muslims read The Koran, but very few Muslims read the Bible, and fewer still (I expect) Christians read the Koran.
In economics, the dominant religion is Neoclassical economics, and the main minority religion is Post-Keynesian economics (Austrian economics is best regarded as a more realistic—and anti-modelling—offshoot of Neoclassical economics). Rendahl and Eggertsson are both Neoclassicals, and they were having a go at me for—so they thought—not knowing of the endogenous growth literature spawned by the work of Paul Romer.
They are wrong, for two reasons.
Firstly, there’s a twist to the religion analogy for economics. Except for a handful of converts, believers in one religion are born, raised and educated in it. But in economics, given that it gets taught in universities rather than churches, the economic religion with the dominant population gets to teach everybody. That is Neoclassical economics, and it dominates teaching at all but a dwindling handful of universities. To become a member of a rival economic religion—be it Post-Keynesian, Austrian, Marxist, Evolutionary, Behavioural, whatever—you first have to endure an inculcation in Neoclassical economics. So, while almost no Neoclassical economist reads non-Neoclassical works, non-Neoclassical economists have at the very least endured several years of undergraduate inculcation in Neoclassical economics.
Secondly, because I’m the ornery bastard who decided to take Neoclassicals on with my book Debunking Economics, I have read a ridiculous volume of Neoclassical works—more, I expect, than most Neoclassical economists themselves. They are incredibly ignorant of even their own religion’s history, because they disdain the history of economic thought so much that they don’t even learn their own history.
Therefore, I am quite aware of Paul Romer’s work on endogenous growth theory (Romer 1990)—and I hold Romer in much higher regard than I do most Neoclassical economists. He is someone who has pushed the boundaries of the Neoclassical paradigm, and he has also been scathingly critical of the mess that Neoclassical economics is in these days, in a brilliant working paper entitled “The Trouble with Macroeconomics” (Romer 2016).
So, why do I nonetheless not take his model of endogenous growth seriously?
There are two primary reasons. The first is its Neoclassical foundation. Romer went to great lengths to situate his work within the school of economics he knew at the time, and that of course was Neoclassical economics, where you have to integrate your element into the fundamentally equilibrium-dominated corpus of Neoclassical economics. Figure 1, which reproduces, the overview paragraph in his paper, is peppered with the concept of “equilibrium”.
Figure 1: An extract from Romer’s endogenous growth model paper
The second is that, at much the same time as Romer was developing Neoclassical endogenous growth theory, the great (but sadly neglected, even by Post-Keynesians) economist Richard Goodwin was developing a Post-Keynesian approach to precisely the same issue (Goodwin 1990).
The insights of the two men were similar, but the framing of their analysis couldn’t have been more different. Whereas Romer shoehorned his insights into the equilibrium fetish of mainstream economics, Goodwin was specifically looking for system that were locally unstable, but globally stable: where the point equilibria were repellers, rather than attractors of the system, but where nonlinearity far from equilibrium meant that the models stayed within realistic bounds:
This, despite the obsession Neoclassicals have with equilibrium, is the real world. Economists have failed to find chaos in economic data (as an ambivalent review of Goodwin’s book pointed out) because of the tests they have applied to it, not because it doesn’t exist. As I have shown in The New Economics: A Manifesto, chaotic dynamics arise from a simple macroeconomic model derived from the definitions for the employment rate, the wages share of GDP, and the private debt to GDP ratio (Keen 2021, pp. 97-100). The resulting model is a relative, if not quite a twin, of Lorenz’s original model of chaotic dynamics in the weather (Lorenz 1963).
Even if this were not true, this is the 21st century, not the 19th. The excuse that the original Neoclassicals had then for doing equilibrium analysis was valid—even if their deductions from it were false:
If we wished to have a complete solution of the problem in all its natural complexity, we should have to treat it as a problem of motion—a problem of dynamics. But it would surely be absurd to attempt the more difficult question when the more easy one is yet so imperfectly within our power.” (Jevons 1888, p. 93)
But in the 21st century, we have software systems which can easily handle far-from-equilibrium behaviour. There is no need to shoe economic phenomena into an equilibrium framework. Here, for example, is Goodwin’s model of endogenous growth in Minsky, the Open Source system dynamics tool I have developed for economic modelling:
Figure 2: Richard Goodwin’s model of innovation-driven growth
Therefore, while I appreciated Romer’s work, he ended up “putting lipstick on a (Neoclassical) pig”. The model still had to generate equilibrium outcomes, and for that reason, the pigs liked it. But they also weren’t particularly willing to put the lipstick on, because, like so many attempts to generalise Neoclassical economics, it forced outcomes that Neoclassicals don’t inherently like. Romer summarized this problem here:
Once the cost of creating a new set of instructions has been incurred, the instructions can be used over and over again at no additional cost … Most models of aggregate growth, even those with spillovers or external effects, rely on price-taking behavior. But once these three premises are granted, it follows directly that an equilibrium with price taking cannot be supported. (Romer 1990, p. S72. Emphasis added)
Since then, there are models in which Neoclassicals want non-price-taking behaviour: DSGE models (which without exception did not predict the Global Financial Crisis) rely on one sector of an economy exhibiting price-setting behaviour (monopoly or “imperfect competition” in the Neoclassical taxonomy), while the rest exhibits the typical “well-behaved” competitive, price-taking behaviour. So, some models in this family might well use Romer’s endogenous growth model to generate the price-setting sector.
But in most of the rest of the Neoclassical lexicon, Romer’s insight is ignored. Models like Nordhaus’s DICE, for example, have a standard production function with exogenous technical change. This is where Eggertsson’s dig at me on Twitter is actually informative:
One wonders what economists he is actually asking. His undergraduate students? Reveals approximately zero knowledge about the literature on growth.
There is indeed a Neoclassical literature on growth—just like there is a Neoclassical literature on wage bargaining, on finance, on utility pricing, etc., and even on climate change—unfortunately (Lenton et al. 2021; Keen 2020). There are economists who spend their whole careers in one of these fields, writing numerous papers extending this aspect of the Neoclassical literature. But these sub-fields of Neoclassical economics never change the core message of Neoclassical economics, because in many ways they are inimical to the core.
Neoclassicals want to believe that (most) markets are competitive (as they define competition), because only if markets are competitive does a “supply curve” exist. They want a production function with constant returns to scale, and incomes based on marginal products, because otherwise their theory of income distribution breaks down. So these literatures exist, and are each voluminous, but the core Neoclassical method is unaffected by them.
Consequently, there is a rich literature of Neoclassical endogenous growth theory models, just like there is a rich literature of crime thrillers, and horror stories, and even love stories, in fiction. But has Neoclassical economics been fundamentally transformed by Romer’s work? Not in the slightest. The vast majority of Neoclassical models don’t apply his insights, just as the vast majority of novels aren’t horror stories.
Lastly, there is a positive contribution of mine that points out a weakness, not just in Neoclassical economics, but in Romer’s attempt to plug a hole in it by providing an endogenous theory of technical change. This is the role of energy in production—something that Rendalh’s contribution to this Twitter debate unintentionally highlighted:
Yes, there is a lot of technological development directed at reducing energy consumption. But when you look at the aggregate global data, there is a quite literally one for one relationship between energy and GWP (Gross World Product, the global extension of Gross Domestic Product):
And even between change in energy and change in GWP:
Why is that? Because “labour without energy is a corpse, while capital without energy is a sculpture” (Keen, Ayres, and Standish 2019, p. 41): energy is an essential input to both labour and capital, without which no work can be done. Therefore, the correct way to think about production is that machines and workers together help transform the energy we find in the environment into useful work.
A Cobb-Douglas Production Function—the mainstay of Neoclassical macroeconomic modelling today—has no role for energy. But if you replace both Capital and Labour with the number of each, times the annual energy consumption of each, times how efficiently they turn that energy into useful work, you get an expression involving the amount of energy that an unskilled worker can put into production (effectively, about 100 watts an hour maximum), times the energy consumption level of the “representative machine” at any given time in history, times the labour and capital components of the Cobb-Douglas Production Function:
This is an alternative explanation to the one that Romer provided for growth in output thanks to technical change: that putting more energy into output has been the primary form that technical progress has taken. Of course it has taken human ingenuity to enable that increased energy consumption, but the manifest cause of rising per capita living standards has been rising energy consumption per head.
This, unlike human ingenuity, can go backwards—and if climate change forces us to consume far less energy, this could collapse precipitously. When this happens—I no longer think that it is a question of “if”—then Romer’s model’s explanation for the resulting decline in GWP would have to be a decline in ingenuity. But the real cause will be a drop in energy consumption per capita.
Eggertsson, Gauti B., and Paul Krugman. 2012. ‘Debt, Deleveraging, and the Liquidity Trap: A Fisher-Minsky-Koo approach’, Quarterly Journal of Economics, 127: 1469–513.
Goodwin, Richard M. 1990. Chaotic economic dynamics (Oxford University Press: Oxford).
Jevons, William Stanley. 1888. The Theory of Political Economy ( Library of Economics and Liberty: Internet).
Keen, Steve. 2020. ‘The appallingly bad neoclassical economics of climate change’, Globalizations: 1-29.
———. 2021. The New Economics: A Manifesto (Polity Press: Cambridge, UK).
Keen, Steve, Robert U. Ayres, and Russell Standish. 2019. ‘A Note on the Role of Energy in Production’, Ecological Economics, 157: 40-46.
Lenton, Timothy, T. J. Garrett, M. Grasselli, Steve Keen, Devrim Yilmaz, and Antoine Godin. 2021. “Economists’ erroneous estimates of damages from climate change.” In. Arxiv.
Lorenz, Edward N. 1963. ‘Deterministic Nonperiodic Flow’, Journal of the Atmospheric Sciences, 20: 130-41.
Romer, Paul. 2016. “The Trouble with Macroeconomics.” In.
Romer, Paul M. 1990. ‘Endogenous Technological Change’, Journal of Political Economy, 98: S71-S102.