Energy as a Driver of Economic Growth

This is the first of what I expect to be several posts in a (not necessarily consecutive) series on energy and economic growth. It fits into the broader question of what drives long-term economic growth.

As mentioned last week, the idea that energy is a central factor, maybe even the central factor, in long-term growth motivated my interest, and later professional work, on the energy topic. But I have since become skeptical that this link is as strong as many researchers think, and in particular, I suspect that hopes that a breakthrough in energy technology will trigger a wider economic boon are misplaced.

Today, we will consider the centrality of energy as a foundational idea in ecological economics, as advanced in Georgescu-Roegen (1971). Then we will review Ayres and Warr (2005), which posits energy as a key factor in long-term growth. Finally, we will review some recent work on the causal link between energy and growth.

Energy in Ecological Economics

Ecological economics is an academic discipline that, as the name suggests, seeks to put ecological principles on an economic basis. Key ideas in the discipline are the interdependence between human civilization and non-human nature and that human civilization is a proper subset of nature, not the other way around. Degrowth, the idea that economic policy should be based on setting a maximum size of the economy that is less than or comparable to the current size, is very much rooted in ecological economics.

I have written about topics related to ecological economics, mostly critically, such as a piece last February on planetary boundaries. To give another example of a difference between ecological economics and ecomodernism, a tradition that I more closely identify with, consider decoupling, which is the notion that it is possible for the economy to grow without an increase in environmental impact. Ecological economists believe that decoupling, to the extent needed, is impossible, and that there is an unavoidable tradeoff between economic activity and environmental quality. Ecomodernists believe that decoupling is possible and that degrowth as a solution to environmental problems is unnecessary and unwise. Warlenius (2023) discusses the disagreement in the context of carbon emissions, taking the ecomodernist position that decoupling between emissions and growth is possible.

The International Society for Ecological Economics and its journal Ecological Economics were founded in 1989 by Herman Daly, a protégé of Nicholas Georgescu-Roegen, although Georgescu-Roegen would later become a critic of Daly’s Steady-State Economics (see Georgescu-Roegen (1977)). Nevertheless, Georgescu-Roegen’s ideas, as expounded in Georgescu-Roegen (1971), would be an important influence on the field.

Georgescu-Roegen’s 1971 work, The Entropy Law and the Economic Process, despite some flaws that we will get to, should be regarded as one of the foundational works in the modern environmental movement that appeared around 1970, alongside Barry Commoner’s The Closing Circle (1971), Paul and Anne Ehrlich’s The Population Bomb (1968) (discussed a few months ago), and The Club of Rome’s The Limits to Growth (1972).

Georgescu-Roegen’s aim is to develop a new economics that is based on the principles of thermodynamics, or “the entropy law” as he puts it. According to Georgescu-Roegen, the dominant paradigm of neoclassical economics is rooted in a mechanistic understanding of reality, which in turns derives from the mechanistic physics of Isaac Newton. Prior to the development of thermodynamics, the neoclassical paradigm of economics had no concept of entropy, and it wrongly assumed that natural resources could be used indefinitely.

With a 19th century understanding of entropy, we know that all economic processes expend usable energy—that is, that they take nature from a state of low entropy to a state of high entropy—and thus they are irreversible. Thus all economic activity “uses up” the low entropy reservoirs on Earth.

Georgescu-Roegen identifies two entropy fluxes that are relevant to humans. There are terrestrial low entropy sources, which include both the energy potential in terrestrial energy sources and the low entropy reservoirs in, for instance, concentrated mineral ores. The supply of these these reservoirs is finite and fixed, and so the greater civilization’s metabolism of them, the sooner they are used up and civilization must come to an end.

The second source is the solar energy flux. This of course is also finite—the Sun will eventually run out—but there is nothing that humans can do to shorten, lengthen, increase, or decrease the solar flux (setting aside speculation about star lifting, as described by Matloff (2017)). Therefore, human activities that harness the Sun, such as agriculture and solar power, do not deplete low entropy reservoirs in the same way that mining and fossil fuel combustion do. Here, one sees the basis for the concepts of extractivism and renewable energy, though both concepts are not fully developed yet.

Georgescu-Roegen’s conclusion is that the ideal course for civilization is to extend its life by reducing population and consumption, so as the maximize the amount of solar energy flux over the lifetime of low entropy reservoirs on Earth.

I can’t discuss all the problems with Georgescu-Roegen’s in every last detail. One of the more glaring is a serious misunderstanding of the concept of entropy, which is a key concept in the book (right in the title). Gillett (2006) points out that the Earth is not a closed system, but rather it is situated between two effectively infinite temperature reservoirs: the Sun and outer space. Thus the surface of the Earth retains low entropy reservoirs after hundreds of millions of years of complex life, and the idea that economic activity will deplete those reservoirs in a time frame relevant to decision-making is senseless. Nevertheless, when you hear slogans such as “thermodynamics forbids endless economic growth”, Georgescu-Roegen (1971) is where that idea comes from.

Energy as a Factor in Growth

Now let us turn to a later paper from ecological economics, Ayres and Warr (2005). This paper seeks to determine the extent to which availability of energy drives economic growth, using a modification of the Solow-Swan growth model.

The Solow-Swan model dates back to the 1950s with a pair of papers, Solow (1956) and Swan (1956). The model expresses economic growth in terms of overall labor supply, deployment of capital, and a residual term that is often identified with technology. The Solow residual can be quite large, and the fact that is was so poorly understood was the biggest shortcoming with the original model.

In the tradition of ecological economics, Ayres and Warr (2005) suggest that the Solow residual can be identified with energy. Or more specifically, with exergy, which is the energy consumption that is useful to the economy. Exergy is calculated by multiplying raw energy consumption by an efficiency factor, which has generally gone up over time. Thus exergy has increased at a faster rate than raw energy. This is important, since either raw energy nor labor nor capital has increase at as fast a rate as overall economic growth, and so no combination with constant returns to scale will be a good fit (I refer the reader who is interested in the mathematical details to the paper).

They also find that it is better to use a more expansive definition of energy than, say, the International Energy Agency or the U.S. Energy Information Administration might use. The definition of Ayres and Warr (2005) entails the usual forms of energy—fossil fuels, uranium, hydropower, etc.—and it also includes the energetic requirements of refined minerals and nonfuel wood products. The difference between the two is not that great, but the more expansive form gives a better fit.

Ayres and Warr (2005) use a linear exponential function production function, which retains the desired assumption of constant returns to scale and fits the data a bit better than the Cobb-Douglas function that is more commonly used; again, consult the paper if you want the mathematical details.

With all that in place, Ayres and Warr (2005) find that with the three factor model, the residual term that one finds when considering only labor and capital mostly disappears when modeling economic growth in the United States from 1900 to 1970. From 1970 to 1998, the last year of data in the paper, the goodness of fit is not quite as good, and a residual term is again needed. They hypothesize two explanations for that. First, it might be that information technology adds a fourth factor of production from 1970 that needs to be modeled. Second, it may be that conservation in response to the oil crises of the 1970s has … ahem … decoupled growth from energy consumption.

Does Energy Cause Growth?

The biggest shortcoming of Ayres and Warr (2005) is that they do not establish causation. They suggest, but do not show, that availability of energy causes the economy to grow; just as plausible an interpretation of their results is that economic growth causes increased energy consumption, or that the two things move together without being causally connected. We will look at a few recent papers to try to discern that.

Granger causality, introduced by Granger (1969), is a statistical test to suggest whether movements in one time series cause movements in another. It works by identifying a correlation between the first series and a lagged version of the second. As Maziarz (2015) argues, Granger causation does not establish causation in an epistemological sense, and even if the test suggests “causation”, one should not assume genuine causation unless a plausible causal mechanism is identified. I would prefer if we didn’t call this “causality”, but that’s what we have. With that out of the way, let’s look at some results.

Tran et al. (2022) examine panel data of 26 countries of the Organisation for Economic Co-operation and Development—generally wealthy countries—of energy consumption and gross domestic product from 1971 to 2014. They find a threshold per-capita GDP of $48,170. Below that threshold, energy consumption is found to Granger cause gross domestic product. Above the threshold, the direction of causality reverses; in the long run, GDP is found to Granger cause energy consumption, while no such relationship is found in the short run. By comparison, per-capita GDP in the United States was just under $83,000 as of 2023. The results suggest that the importance of energy as a driver of economic growth diminishes for high levels of wealth.

Alola, Bekun, Sarkodie (2019) examine several variables related to ecological impact in 16 European Union countries from 1997 to 2014. They find that gross domestic product Granger causes nonrenewable energy consumption, but not the reverse. They also find a mutually causal relationship between renewable energy consumption and GDP.

Perera et al. (2024) consider the share of renewable and nonrenewable energy in a country's mix, and they examine 152 countries with panel data consisting of the share of renewable and nonrenewable energy and of gross domestic product. The countries are divided into four groups: developed, developing, least developed, and transitional. Among the four groups, the authors found that the share of renewable energy positively Granger causes economic growth in transitional countries (generally former republics of the Soviet Union), and no statistically significant relationship was found between the variables in the other country groups, as well as the world as a whole. The authors find that more statistically significant relationships hold for many individual countries.

These studies are not conclusive, and there is much more that I haven’t looked at yet for lack of time. But based on both Ayres and Warr (2005) and the three papers above, an argument can be made that, while greater access to energy supplies may have been a driver of growth in the past, its importance diminishes for wealthy countries.

For today, I’ll leave the last word to Stern (2011). He argues that, while energy has historically been an important driver of growth, its importance is declining, also using Ganger causality analysis. A simple explanation for this may be that the cost of energy, as a share of GDP, has generally been going down, and the cost as share of GDP is a good proxy for a variable’s importance for growth. The world on the eve of the Industrial Revolution was starved for energy and greatly benefited when it became available in great quantities. The world today experiences energy abundance, and so we should not expect, for instance, a revolution in enhanced geothermal drilling, small modular reactors, or fusion power to catalyze the same kind of change the world saw 200 years ago.

Quick Hits

Relevant to today’s topic, the International Energy Agency recently published an analysis of energy and data centers. both on the energy consumption of the data centers themselves and the capability of artificial intelligence to make the energy system more efficient. There is more of an energy abundance tone to the report than anything I have seen from the IEA in a while.

Senators Bill Cassidy (R-LA) and Lindsay Graham (R-SC) recently introduced the Foreign Pollution Fee, with this two-pager. This bill has been introduced in the past. It is essentially a border adjustment, in that it would impose a tariff on imported goods based on their monetized pollution intensity. Here is Citizens Climate Lobby in favor of the bill, and the American Action Forum against it. The bill sounds like a carbon tax, which I support, but since it would apply only to imports, and no commensurate carbon pricing is in place for domestic production, it doesn’t look like much more than protectionism.

Ted Nordhaus’ piece Nuclear Underpants Gnomes, so named to reference a meme from a Southpark episode, is critical of several simplistic solutions that are offered by proponents of nuclear power. “Underpants gnome” has become a byword for half-baked business ideas that will supposedly work by mechanisms that are not clear the listener and probably not to the speaker either. In the case of nuclear power, the simplistic, half-baked ideas are abolishing the Nuclear Regulatory Agency, building a bunch of light water reactors (hey, it worked for France in the 1980s, right?), or not building wind and solar power—presumably so nuclear power will be built instead?