Long-run Benefits of Electrification Paper Accepted for Publication

My paper with Suryadeepto Nag on the long-run benefits of electrification has finally been accepted for publication in the Energy Journal! It has taken us three years since we put out a working paper. These are all the submissions we tried:

• AEJ-AE: Submitted 26Jul23. Desk-rejected 28Jul23. 
• JDE: Submitted 31Jul23. Rejected 21Dec23. 
• JAERE: Submitted 9Mar24. Desk-Rejected 14Mar24. 
• World Development: Submitted 15Mar24. R&R 17May24. Resubmitted 29Jul24. Rejected 11Feb25. 
• Energy Economics: Submitted 25Feb25. Rejected 17May25.
• Energy Journal: Submitted 24Sep25. R&R 29Dec26. Resubmitted 7Apr26.Accepted 8Jul26

OK, so our first three submissions were apparently too high ranked, though we got reviewed at JDE. But really, I think if Energy Journal accepted, either of WD or Energy Economics should have.

Are the Benefits of Electrification Realized Only in the Long Run? Evidence from Rural India

 

I have a new working paper coauthored with my master's student Suryadeepto Nag on the impact of rural electrification in India. Surya did his master's at IISER in Pune with me as his supervisor. He visited Canberra over the last Southern Summer. This paper is based on part of Surya's thesis.

The effect of providing households with access to electricity has been a popular research topic. It's still not clear how large the benefits of such interventions are. Is electricity access an investment that generates growth? Or is it more of a consumption good that growing economies can afford? Researchers have used traditional econometric methods on secondary data (observational studies) and also carried out field experiments, such as randomized controlled trials (RCTs), to try to answer this question.

Experimental methods have generally found smaller and less statistically significant results than observational studies have. Is this because experiments are more rigorous? Or because observational studies usually measure impacts over a longer period of time? It's likely that it takes time for people to make use of a new electricity connection. They will need to save and buy appliances. Effects on children's education will take an especially long time to come to fruition.

We carry out a meta-analysis of 16 studies previously reviewed by Bayer et al. (2020):

We assigned each positive impact (for example on income or on education) found in a study the score of 1 and each negative impact a -1 and then averaged over all the impacts. The graph shows this "positiveness of impact" compared to the time households had been connected to electricity. While observational studies found more positive impacts than experimental studies, there is also a positive correlation between duration of connection and positiveness of impact (and between duration of connection and being an observational study). Regression analysis shows that only duration of connection is statistically significant. 

But this small sample of studies can't be that conclusive, so we then carry out our own analysis to test whether impacts increase over time.

Using three waves of Indian household surveys from 1994-95, 2004-5, and 2011-12, we quantify the impacts of short-term (0-7 years) and long-term (7-17 years) electricity access on rural household well-being. These surveys tracked the same households over time. We don't know exactly when a household was connected, just whether it was already connected in 1994-95 or whether it got connected between the other surveys. We do know when villages were connected.

We use a difference in differences regression that is weighted using "inverse propensity scores". This is supposed to compensate for the fact that households are not actually connected randomly to the grid. If, for example, poor households are less likely to get connected, we overweight them in the sample. In our main analysis, we exclude households that were already connected in 1994-95 so that the control group only includes households that were not connected by 2011-12.*

We find that long-term electricity access increases per capita consumption and education, and reduces the time spent by women on fuel collection (compared to the control group). The effect of short-term connection is smaller and statistically insignificant. We find no significant effects on agricultural income, agricultural land holding, and kerosene consumption. 

Here is our main table of results:

The long-term impact on consumption is really very big – 18 percentage points more than the control group over a 7 year period. The effect on education is 0.4 of a year relative to the control group.

We did some robustness tests – using different weights and including the households connected before 1994-95 as "very long-term connections". The results roughly hold up, though the weighting isn't ideal in either case.

We think our results show that experimental studies really need longer term follow-ups before coming to conclusions.

* The recent research on differences in differences shows that many past studies used inappropriate control groups.

China is pumping out carbon emissions as if COVID never happened. That’s bad news for the climate crisis

David Stern, Crawford School of Public Policy, Australian National University and Khalid Ahmed, Australian National University

Carbon emissions from China are growing faster now than before COVID-19 struck, data show, dashing hopes the pandemic may have put the world’s most polluting nation on a new emissions trajectory.

We compared emissions in China over the first four months of 2019 – before the pandemic – and 2023. Emissions rose 10% between the two periods, despite the pandemic and China’s faltering economic recovery. Power generation and industry are driving the increase.

Under the Paris Agreement, China has pledged to ensure carbon emissions peak by 2030 and reach net zero emissions by 2060. Our analysis suggests China may struggle to reach these ambitious goals.

Many believed the economic recovery from COVID would steer global development towards a less carbon-intensive footing. But China’s new path seems to be less sustainable than before. That’s bad news for global efforts to tackle climate change.

 

China has pledged to ensure carbon emissions peak by 2030 – but it’s heading in the opposite direction. Olivia Zhang/AP

An alarming trend in emissions

The COVID pandemic curbed greenhouse gas emissions in 2020, largely due to a drop in passenger travel. This led to hopes of a “green” economic recovery in which government stimulus spending would be invested into climate-friendly projects, to ensure a longer-term slowing of growth in emissions.

Some researchers examined the trends in China’s emissions up to 2019 and predicted the nation’s emissions would peak by 2026. Others have said the peak will occur even earlier, in 2025.

But unfortunately, it seems those predictions were too optimistic.

We examined data from Carbon Monitor, which provides science-based estimates of daily CO₂ emissions across the world. We compared emissions data from January to April 2019 (which represents typical pre-pandemic conditions in China) with the corresponding months in 2023. This period followed the removal of most COVID-related restrictions in China – such as testing requirements and quarantine rules – which essentially restored the country’s economy to business-as-usual.

We found average daily carbon emissions increased substantially between the two periods. In the first four months of 2019, China’s transport, industry, energy and residential sectors together emitted an average 28.2 million tonnes of CO₂ a day. In the first four months of 2023, daily emissions from those sectors were an average 30.9 million tonnes.

Emissions from the residential and transport sectors didn’t change much. This is mildly good news – it’s better than emissions going up. But these are the two smallest sectors, together accounting for only 18% of China’s emissions.

Rather, the increase was driven by emissions from China’s industrial and energy sectors. Average daily emissions from industry rose between 2019 and 2023 by 1.1 million tonnes or 11%. From energy, which includes electricity generation, they rose by 1.75 million tonnes or 14%.

Energy production from solar and wind in China did increase substantially between the two periods. But the growth was outweighed by electricity generated from fossil fuels.

Graph showing energy generation mix in China in the first four months of both 2019 and 2023. National Bureau of Statistics of China

Separate data show the growth of coal production in China has accelerated. In the two years prior to the pandemic, coal production variously fell or only grew slightly. But coal production grew during the pandemic, and this has continued. In the year to April 2023, coal production increased by about 5%.

While coal’s share of energy consumption fell substantially from 2007 to 2019, it has changed little since then. That’s mainly because energy use is growing fastest in the electricity sector, which remains dominated by coal.

The global picture

Emissions in many developed countries have fallen in recent years due to government policies, slow economic growth, and the shift from coal to natural gas.

Developing nations increasingly dominate global emissions. China might be expected to be a leader on the clean energy shift among developing countries – in part because it produces much less oil than it consumes. That means its energy supply is not secure, giving it an incentive to find alternative sources of power.

There’s another reason why China should be a trailblazer on emissions reduction. China is the world’s biggest emitter – so a percentage reduction in emissions there leads to far fewer tonnes of CO₂ in the atmosphere than if a smaller country reduced emissions by the same percentage. And, partly because China’s population and economy are so big, it stands to benefit more than any country in the world from a more stable global climate.

But as we’ve outlined, China’s trajectory is by no means world-leading. What’s more, moves by China on the international stage suggest it’s becoming less cooperative in climate negotiations than in recent years. We saw this at the COP27 global climate conference in Egypt late last year, when China did not join a pledge to curb methane emissions and refused to provide financial support to developing nations vulnerable to climate change.

The potential for cooperation on climate policy is being reduced further by ongoing tensions between China and the United States. All this serves to cast doubt on China following through on its Paris pledges – and certainly, on any chance its emissions will peak in the next two years.

David Stern, Professor, Crawford School of Public Policy, Australian National University and Khalid Ahmed, Visiting Fellow, Australian National University

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Confidence Intervals for Recursive Journal Impact Factors

I have a new working paper coauthored with Johannes König and Richard Tol. It's a follow up to my 2013 paper in the Journal of Economic Literature, where I computed standard errors for simple journal impact factors for all economics journals and tried to evaluate whether the differences between journals were significant.* In the new paper, we develop standard errors and confidence intervals for recursive journal impact factors, which take into account that some citations are more prestigious than others, as well as for the associated ranks of journals. We again apply these methods to the all economics journals included in the Web of Science.

Recursive impact factors include the popular Scimago Journal Rank, or SJR, and Clarivate's Article Influence score. We use Pinski and Narin's invariant method, which has been used in some rankings of economics journals. 

As simple impact factors are just the mean citations an article published in a journal in a given period receives in a later year, it is easy to compute standard errors for them using the formula for the standard error of the mean. But the vector of recursive impact factors is the positive eigenvector of a matrix and its variance does not have a simple analytical form.

So, we use bootstrapping to estimate the distribution of each impact factor. Taking all 88,928 articles published in 2014-18 in the economics journals included in the Web of Science, we resample from this dataset and compute the vector of recursive impact factors from the new dataset.** Repeating this 1,000 times we pick the 2.5% or 97.5% range of values for each journal to get a 95% confidence interval:

95% confidence intervals of the recursive impact factor, arithmetic scale (left axis) and logarithmic scale (right axis).

The graph repeats the same data twice with different scales so that it's possible to see some detail for both high- and low-ranked journals. Also, notice that while the confidence intervals for the highest ranked journals are quite symmetric, confidence intervals become increasingly asymmetric as we go down the ranks. 

The top ranked journal, the Quarterly Journal of Economics, clearly stands out above all others. The confidence intervals of all other journals overlap with those of other journals and so the ranks of these journals are somewhat uncertain.*** So, next we construct confidence intervals for the journals' ranks.

It turns out that there are a few ways to do this. We could just construct a journal ranking for each iteration of the bootstrap and then derive the distribution of ranks for each individual journal across the 1,000 iterations. Hall and Miller (2009), Xie et al. (2009), and Mogstad et al. (2020) show that this procedure may not be consistent when some of the groups (here journals) being ranked are tied or close to tied. The corrected confidence intervals are generally broader than the naive bootstrap approach.

We compute confidence intervals for ranks using the simple bootstrap, the Xie et al. method, and the Mogstad et al. method:

 

95% confidence intervals of the rank based on the recursive impact factor. The inner intervals are based on Goldstein’s bootstrap method, the middle intervals use Xie’s correction to the bootstrap, and the outer intervals follow Mogstad’s pairwise comparison.

The simple bootstrap under-estimates the true range of ranks, while it seems that the Mogstad et al. method might be overly conservative. On the other hand, Xie et al.' s approach depends on choosing a couple of "tuning parameters".  

All methods agree that the confidence interval of the rank of the Quarterly Journal of Economics only includes one. Based on the simple bootstrap, the remainder of the "Top-5'' journals are in the top 6 together with the Journal of Finance, while the Xie et al. method and the Mogstad et al. methods generally broaden estimated confidence intervals, particularly for mid-ranking journals. All methods agree that most apparent differences in journal quality are, in fact, mostly insignificant. We think that impact factors, whether simple or recursive should always be published together with confidence intervals.

* The latter exercises were a bit naive. As pointed out by Horrace and Parmeter (2017), we need to account for the issue of multiple comparisons. 

** Previous research on this topic resampled at the journal level, missing most of the variation in citation counts.

*** Overlapping confidence intervals don't neccessarily mean that there is no signficant difference between two means. Goldstein and Harvey (1995) show that the correct confidence intervals for such a test of the difference between two means are narrower than the conventional 95% confidence intervals. On the other hand, for multiple comparisons we would want wider confidence intervals.

Estimating the Effect of Physical Infrastructure on Economic Growth

I have a new working paper coauthored with Govinda Timilsina of the World Bank and my PhD student Debasish Das. It is a panel data study of the effect of various forms of infrastructure on the level of GDP. 

Compared to existing studies, we use more recent data, include new types of infrastructure such as mobile phones, and provide separate estimates for developing and developed countries. We find larger effects than most previous studies. We also find that infrastructure has a larger effect in more recent years (1992-2017) than in earlier years (1970-1991), and the effects of infrastructure are higher in developing economies than in industrialized economies. The long-run effects seem to be much larger than the initial impact. We also tried to estimate the effect of infrastructure on the rate of economic growth. Controlling for the initial level of GDP per worker we found a null result. So, we can't say that having more infrastructure means a more rapid rate of economic growth.

Getting good quality data that is comparable across countries is really a problem in this area of research. Many types of infrastructure only have data available for a few years. The ones that have more panel-like data often suffer from differences in definition across countries – such as what is a road or a motorway – or unexplained jumps in individual countries. So, our results are subject to a lot of measurement error.

Our main analysis uses data on five types of infrastructure – roads, railways, electric generation capacity, fixed line telephones, and mobile telephones*:

Following some previous research, we aggregate the individual types of infrastructure using principal component analysis. We use two principal components. One factor seems to be related to transport infrastructure and the other to electricity and telecommunications. Still, we can recover estimates of the effect of each individual type of infrastructure.

Also following some previous research, we use the Pooled Mean Group estimator to estimate a dynamic panel regression model. This allows us to test for the weak exogeneity of the explanatory variables, allowing us to give the results a somewhat causal interpretation.

The table shows the percentage change in GDP per worker for a 1% change in each infrastructure type. Getting standard errors for these estimates would be rather tricky.** Interestingly, the PMG estimates are mostly much larger than the static fixed effects estimates. Static fixed effects can be expected to converge to a short-run estimate of the effects while PMG should be a better estimate of long-run effects. Fixed effects also tends to inflate the effects of measurement error. 

Maybe the most innovative thing in the paper is that we plot the impulse response functions of GDP with respect to a 1% increase in each of the two main types of infrastructure:

PC1 is electricity and communications and PC2 transport infrastructure.*** Long-run effects of infrastructure are much larger than the short-run effects. In the short run, transport infrastructure even has a negative impact.

* Note that the graphs show the country means of these variables, while we actually use the deviations from those means over time in each country. 

** We only estimate the GDP-infrastructure relationship, but I think we would need time series models for each of the explanatory variables in order to sample from those models' residuals in a bootstrapping procedure. Bootstrapping is needed because we first carry out the principal components analysis and then estimate the PMG model in a second stage. These elasticities are combinations of the parameters from those two models.

*** We could get a confidence interval for these impulse response functions if we assume that the explanatory variables in the PMG model are deterministic as this analysis assumes...

The Environmental Kuznets Curve: 2021 Edition

The second encyclopedia chapter. First one is here.

Introduction 

The environmental Kuznets curve (EKC) is a hypothesized relationship between various indicators of environmental degradation and countries’ gross domestic product (GDP) per capita. In the early stages of a country’s economic development, environmental impacts and pollution increase, but beyond some level of GDP per capita (which will vary for different environmental impacts) economic growth leads to environmental improvement. This implies that environmental impacts or emissions per capita are an inverted U-shaped function of GDP per capita, whose parameters can be statistically estimated. Figure 1 shows a very early example of an EKC. A large number of studies have estimated such curves for a wide variety of environmental impacts ranging from threatened species to nitrogen fertilizers, though atmospheric pollutants such as sulfur dioxide and carbon dioxide have been investigated most. Panayotou (1993) was the first to call this relationship the EKC, where Kuznets refers to the similar relationship between income inequality and economic development proposed by Nobel Laureate Simon Kuznets and known as the Kuznets curve. The EKC can be seen as an empirical version of the interpretation of sustainable development as the idea that development is not necessarily damaging to the environment and, also that poverty reduction is essential to protect the environment (World Commission on Environment and Development, 1987).

Figure 1. An Environmental Kuznets Curve 

The EKC has been the dominant approach among economists to modeling ambient pollution concentrations and aggregate emissions since Grossman and Krueger (1991) introduced it in an analysis of the potential environmental effects of the North American Free Trade Agreement. The EKC also featured prominently in the 1992 World Development Report published by the World Bank and has since become very popular in policy and academic circles and is even found in introductory economics textbooks. 

Critique 

Despite this, the EKC was criticized almost from its beginning on empirical and policy grounds, and debate continues. It is undoubtedly true that some dimensions of environmental quality have improved in developed countries at the same time that they have become richer. City air and rivers in these countries have become cleaner since the mid-20th Century and, in some countries, forests have expanded. Emissions of some pollutants such as sulfur dioxide have clearly declined in most developed countries in recent decades. But there is more mixed evidence for other pollutants such as carbon dioxide. Carbon emissions have fallen in the last 40 years in some developed countries such as the United Kingdom or Sweden, while they have increased in others such as Australia or Japan. There is also evidence that emerging countries take action to reduce severe pollution. For example, Japan cut sulfur dioxide emissions in the early 1970s following a rapid increase in pollution when its income was still below that of the developed countries (Stern, 2005) and China has also acted to reduce sulfur emissions in recent years. As further studies were conducted and better data accumulated, many of the econometric studies that supported the EKC were found to be statistically fragile. 

Initially, many understood the EKC to imply that the best way for developing countries to improve their environment was to get rich (e.g. Beckermann, 1992). This alarmed others (e.g. Arrow et al., 1995), as while this might address some issues like deforestation or local air pollution, it would likely exacerbate other environmental problems such as climate change. Even if there is an EKC for per capita impacts, environmental impacts would increase for a very long time if the majority of the population is on the rising part of the curve and/or the population is also growing (Stern et al., 1996). 

Explanations 

The existence of an EKC can be explained either in terms of deep determinants such as technology and preferences or in terms of scale, composition, and technique effects, also known as “proximate factors”. Scale refers to the effect of an increase in the size of the economy, holding the other effects constant, and should increase environmental impacts. The composition and technique effects must outweigh this scale effect for pollution or other environmental impacts to fall in a growing economy. The composition effect refers to the economy’s mix of different industries and products, which differ in pollution intensities. Finally, the technique effect refers to the remaining change in pollution intensity. This will include contributions from changes in the input mix, for example substituting natural gas for coal; changes in productivity that result in less use, ceteris paribus, of polluting inputs per unit of output; and pollution control technologies that result in less pollutant being emitted per unit of polluting input. 

Over the course of economic development, the mix of energy sources and economic outputs tends to evolve in predictable ways. Economies start out mostly agricultural and the share of industry in economic activity first rises and then falls as the share of agriculture declines and the share of services increases. We might expect the impacts associated with agriculture, such as deforestation, to decline, and naively expect the impacts associated with industry, such as pollution, would first rise and then fall. However, the absolute size of industry rarely does decline, and it is improvement in productivity in industry, a shift to cleaner energy sources, such as natural gas and hydro-electricity, and pollution control that eventually reduce some industrial emissions. On the other hand, offshoring of pollution probably plays only a small role in cutting emissions in developed economies (Kander et al., 2015). 

Static theoretical economic models of deep determinants, that do not try to also model the economic growth process, can be summarized in terms of two parameters: The elasticity of substitution between dirty and clean inputs, which summarizes how difficult it is to cut pollution; and the elasticity of the marginal utility of consumption with respect to consumption, which summarizes how hard it is to increase consumer well-being with more consumption (Pasten and Figeroa, 2012). It is usually assumed that these consumer preferences are translated into policy action. Pollution is then more likely to increase as the economy expands, the harder it is to substitute other inputs for polluting ones and the easier it is to increase consumer well-being with more consumption. If these parameters are constant, then either pollution rises or falls with economic growth. Only if they change over time will pollution first rise and then fall. The various theoretical models can be classified as ones where the EKC is driven by changes in the elasticity of substitution as the economy grows or models where the EKC is primarily driven by changes in the elasticity of marginal utility. 

Dynamic models that model the economic growth process alongside changes in pollution are harder to classify. The Green Solow Model developed by Brock and Taylor (2010) explains changes in pollution as a result of the competing effects of economic growth and a constant rate of improvement in pollution control. Fast growing middle-income countries, such as China, then having rising pollution, and slower growing developed economies, falling pollution. An alternative model developed by Ordás Criado et al. (2011) also suggests that pollution rises faster in faster growing economies but that there is also convergence so that countries with higher levels of pollution are more likely to reduce pollution faster than countries with low levels of pollution. 

Recent Empirical Research 

Recent empirical research builds on these dynamic models to paint a subtler picture than early EKC studies did (Stern, 2017). We can distinguish between the impact of economic growth on the environment and the effect of the level of GDP per capita, irrespective of whether an economy is growing or not, on reducing environmental impacts. We can also distinguish between the effects of economic growth and the simple passage of time. Economic growth usually increases environmental impacts, but the size of this effect varies across impacts and the impact of growth often declines as countries get richer. However, richer countries are often likely to make more rapid progress in reducing environmental impacts. In econometric terms, the time effect – the change in emissions if economic growth is zero – may be higher in richer countries. Rapid growth in middle-income countries, such as China or India, is more likely to overwhelm the time effect in those countries as suggested by Brock and Taylor (2010).

Finally, there is often convergence among countries, so that those that have relatively high levels of impacts reduce them faster or increase them slower than countries with low levels of impacts. These combined effects explain more of the variation in pollution emissions or concentrations than either the classic EKC model or models that assume that either only convergence or growth effects alone are important. Therefore, while being rich means a country might do more to clean up its environment, getting rich is likely to be environmentally damaging. 

References 

Arrow, K., Bolin, B., Costanza, R., Dasgupta, P., Folke, C., Holling, C. S., Jansson, B.-O., Levin, S., Mailer, K.-G., Perrings, C., Pimental, D., 1995. Economic growth, carrying capacity, and the environment. Science 268, 520–521. 

Beckermann, W., 1992. Economic growth and the environment: whose growth? Whose environment? World Development 20, 481–496. 

Brock, W. A.,Taylor, M. S., 2010. The green Solow model. Journal of Economic Growth 15, 127–153. 

Grossman, G. M., Krueger, A. B., 1991. Environmental impacts of a North American Free Trade Agreement. NBER Working Papers 3914. 

Kander, A., Jiborn, M., Moran, D. D., Wiedmann T. O., 2015. National greenhouse-gas accounting for effective climate policy on international trade. Nature Climate Change 5, 431–435. 

Ordás Criado, C., Valente, S., Stengos, T., 2011. Growth and pollution convergence: Theory and evidence. Journal of Environmental Economics and Management 62, 199–214. 

Panayotou, T., 1993. Empirical tests and policy analysis of environmental degradation at different stages of economic development. Working Paper, Technology and Employment Programme, International Labour Office, Geneva, WP238. 

Pasten, R., Figueroa, E., 2012. The environmental Kuznets curve: A survey of the theoretical literature. International Review of Environmental and Resource Economics 6, 195–224. 

Stern, D. I., 2005. Beyond the environmental Kuznets curve: Diffusion of sulfur-emissions-abating technology. Journal of Environment and Development 14(1), 101–124. 

Stern, D. I., 2017. The environmental Kuznets curve after 25 years. Journal of Bioeconomics 19, 7–28.

Stern, D. I., Common, M. S., Barbier, E. B., 1996. Economic growth and environmental degradation: the environmental Kuznets curve and sustainable development. World Development 24, 1151–1160. 

World Commission on Environment and Development, 1987. Our Common Future. Oxford: Oxford University Press.  

Energy and Development

The first of two book chapters for Elgar encyclopedias I recently wrote.

What is the Role of Energy in Economic Activity?

The economic system must operate within the constraints determined by the laws of physics and human knowledge of technology. Production, including household production, requires energy to carry out work to convert materials into desired products and to transport raw materials, goods, and people. The second law of thermodynamics implies that energy cannot be recycled and that there are limits to how much energy efficiency can be improved. Therefore, energy is an essential factor of production, and continuous supplies of energy are needed to maintain existing levels of economic activity as well as to grow and develop the economy (Stern, 1997). The first law of thermodynamics states that energy cannot be created and so energy (and matter) must be extracted from the environment. Also, energy must be invested in order to capture useful energy (Hall et al., 1986). Before the Industrial Revolution, economies depended on energy from agricultural crops and wood as well as a smaller amount of wind and waterpower, all of which are directly dependent on the sun (Kander et al., 2015). This is still largely the case in the rural areas of the least developed countries. While solar energy is abundant and inexhaustible, it is very diffuse compared to concentrated fossil fuels. This is why the shift to fossil fuels in the Industrial Revolution relaxed the constraints on energy supply and, therefore, on production and growth (Wrigley, 1988).

How Does Energy Use Change with Economic Development?

Figure 1 shows that energy use per capita increases with GDP per capita, so that richer countries tend to use more energy per person than poorer countries. The slope of the logarithmic regression line implies that a 1% increase in income per capita is associated with a 0.8% increase in energy use per capita. As a result, energy intensity – energy used per dollar of GDP – is on average lower in higher income countries. These relationships have been very stable over the last several decades (Csereklyei et al., 2016). Energy intensity in today’s middle-income countries is similar to that in today’s developed countries when they were at the same income level (van Benthem, 2015).

Figure 1. GDP and Energy Use per Capita 2018

Energy intensity has also converged across countries over time, so that countries that were more energy intensive in the 1970s tended to reduce their energy intensity by more than less energy intensive countries, and the least energy intensive countries often increased in energy intensity. Though data are limited to fewer and fewer countries as we go back further in time, these relationships also appear to hold over the last two centuries – energy use increased, energy intensity declined globally, and countries converged in energy intensity (Csereklyei et al., 2016). Though data is even more limited, it seems that the share of energy consumption expenditure and production costs also declines as countries develop (Csereklyei et al., 2016; Burke et al., 2018).

The mix of fuels used changes over the course of economic development. Figure 2 shows the average mix of energy sources in each of five groups of countries ordered by income per capita in 2018. In the lowest income countries in the sample (approximately below $5,000 per capita in 2017 purchasing power parity adjusted dollars), traditional use of biomass such as wood and agricultural waste dominates and oil use for transportation as well as electricity generation and other uses is the second most important energy source. As we move to richer countries, the relative role of biomass declines radically, and first oil and then natural gas and primary electricity increase in importance. Note that biomass use per capita in the richest quintile (above $40,000 per capita) is actually greater than in the lowest quintile, as total energy use increases with income. The ways in which this biomass is used will of course be quite different. Higher quality fuels are those that provide more economic value per joule of energy content by being converted more efficiently, being more flexible or convenient to use, and by producing less pollution. We would expect that lower income households would be more willing to tolerate the inconvenience and pollution caused by using lower quality fuels to produce energy services. So as household income increases, we would expect households to gradually ascend an “energy ladder” by consuming higher quality fuels and more total energy. Recent studies often find a more ambiguous picture where multiple fuels are used simultaneously as modern fuels are added to the use of traditional fuels (Gregory and Stern, 2014). 

Figure 2. Fuel Mix and Development 2018

 

In 2016, approximately one billion people remained without access to electricity at home (International Energy Agency, 2017). Around 85% of these people lived in rural areas. There has been rapid progress in electrification in recent years with both grid expansion and the spread of off-grid systems (Burke et al., 2018; Lee et al., 2020). Due to the complexity and costs of electricity-sector management and constrained and weak institutions, power supply is usually less reliable in developing countries than in developed countries (Figure 3) and electricity theft is also more common (Burke et al., 2018). Best and Burke (2017) found that countries with higher levels of government effectiveness have achieved greater progress in providing access to reliable electricity. Industry and other electricity consumers, therefore, often rely on self-generation of electricity, but this is a costly solution (Fingleton-Smith, 2020). 

 Figure 3. Electricity Reliability and Development 2017

 

Does Energy Use Drive Economic Growth?

Economic growth refers to the process that results in increasing GDP per capita over time while development refers to a broader range of indicators including health, education, and other dimensions of human welfare. However, GDP per capita is highly, although not perfectly, correlated with broader development measures (Jones and Klenow, 2016) and so it is worth considering what the role of energy is in economic growth.

Mainstream economic growth models largely ignore the role of energy in economic growth and focus on technological change as the long-run driver of growth. On the other hand, there is a resource economics literature that investigates whether limited energy or other resources could constrain growth. By contrast, many ecological economists believe that energy plays the central role in driving growth and point to the switch traditional energy sources to fossil fuels as the cause of the industrial revolution (Stern, 2011). 

To reconcile these opposing views, Stern and Kander (2012) modified Solow’s neoclassical growth model (Solow, 1956) by adding an energy input that has low substitutability with capital and labor. Their model also breaks down technological change into those innovations that directly increase the productivity of energy– energy-augmenting technical change and those that increase the productivity of labor – labor-augmenting technical change. In this model, when energy is superabundant the level of the capital stock and output are determined by the same functions of the same factors as in the Solow model. But when energy is relatively scarce, the size of the capital stock and the level of output depends on the level of energy supply and the level of energy-augmenting technology. Therefore, in the pre-industrial era and possibly when energy was scarce – and possibly in developing countries today – the level of output was determined by the supply of energy and the level of energy augmenting technology. Until the industrial revolution, output per capita was generally low and economic growth was not sustained (Maddison, 2001). After the industrial revolution, as energy became more and more abundant, the long-run behavior of the model economy becomes more and more like the Solow growth model. If this model is a reasonable representation of reality, then mainstream economists are not so wrong to ignore the role of energy in economic growth in developed economies where energy is abundant, but their models have limited applicability to both earlier historical periods and possibly to today’s developing countries. McCulloch and Zileviciute (2017) find that electricity is often cited as a binding constraint on growth in the World Bank’s enterprise surveys. Energy is expensive relative to wages in developing countries. The price of oil is set globally, and the share of electricity in costs or expenditures can be very high in middle income countries (Burke et al., 2018).

Electricity and Development

Access to energy and electricity, in particular, is a key priority for policymakers and donors in low-income countries. For example, the United Nations’ Sustainable Development Goal 7 targets universal access to modern energy by 2030. Electrification can allow poor households to have easy access to lighting for evening chores or studying and power for phone charging and for a range of new small business activities, both on and off the farm (Lee et al., 2020). Electricity access allows a reallocation of household time, especially for women, away from obtaining energy, for example by collecting firewood, and towards more productive activities. Electricity could also provide health benefits by allowing deeper wells, refrigeration, reduced exposure to smoke etc. (Toman and Jemelkova, 2003).

The micro-level effect of electrification is a growing area of empirical research (Lee et al., 2020). While micro studies typically suggest positive impacts of electrification on income and other development outcomes, more recent quasi-experimental approaches such as randomized controlled trials typically find a smaller impact for electrification than earlier studies did (Lee et al., 2020). Estimates of the effect of electricity infrastructure on economic growth are typically small. One of the best studies (Calderón et al., 2015) estimates the elasticity of GDP with respect to electricity generation capacity as 0.03 (Burke et al., 2018).

Lee et al. (2020) argue that providing poor households with access to electricity alone is not enough to improve economic and noneconomic outcomes in a meaningful way. Complementary inputs are needed, which will accumulate very slowly. Imagination and role models are also important in understanding how to exploit electricity to develop businesses (Fingleton-Smith, 2020). When electricity becomes available in rural areas of sub-Saharan Africa, it is often not used to power agricultural or other productive activities (Bernard, 2012). Institutions are also vital for attaining broad-based benefits from electricity in developing countries. Many developing countries have reformed their electricity sectors during the last few decades, mostly towards market liberalization and corporatization. These efforts have only been partially successful in promoting efficient pricing and greater electricity access (Jamasb et al., 2017). Studies assessing the economic effects of these reforms are scarce. The effects on economic growth seem positive, while the effects on poverty are mixed (Jamasb et al., 2017). In this context, technology transfer and development finance will be critical for increasing the use of electricity in developing countries (Madlener, 2009).

Burke et al. (2018) examined electrification success stories - countries that, from a low level of economic development, have now achieved near-universal electricity access as well as relatively high levels of electricity use. These countries are South Korea, China, Thailand, Vietnam, Egypt, and Paraguay. The first four are well-known development success stories too. Paraguay has abundant hydroelectricity and both Paraguay and Egypt have had relatively strong economic growth. Egypt has been less successful in providing a reliable electricity supply. The most successful countries in increasing access in Sub-Saharan Africa have been South Africa and Ghana, which both suffer from unreliable electricity, which constrains economic activity.

References

Bernard, T., 2012. Impact analysis of rural electrification projects in Sub-Saharan Africa. World Bank Research Observer 27(1): 33–51.

Best, R., and P. J. Burke, 2017. The importance of government effectiveness for transitions toward greater electrification in developing countries. Energies 10(9): 1247.

Burke P. J., D. I. Stern, and S. B. Bruns, 2018. The impact of electricity on economic development: a macroeconomic perspective. International Review of Environmental and Resource Economics 12(1): 85–127.

Calderón, C., E. Moral-Benito, and L. Servén, 2015. Is infrastructure capital productive? A dynamic heterogeneous approach. Journal of Applied Econometrics 30: 177–198.

Csereklyei Z., M. d. M. Rubio Varas, and D. I. Stern, 2016. Energy and economic growth: The stylized facts. Energy Journal 37(2): 223–255.

Fingleton-Smith, E., 2020. Blinded by the light: The need to nuance our expectations of how modern energy will increase productivity for the poor in Kenya. Energy Research & Social Science 70: 101731.

Gregory, J. and D. I. Stern, 2014. Fuel choices in rural Maharashtra. Biomass and Bioenergy 70: 302–314.

Hall, C. A. S., C. J. Cleveland, and R. K. Kaufmann, 1986. Energy and Resource Quality: The Ecology of the Economic Process. New York: Wiley Interscience.

International Energy Agency, 2017. Energy Access Outlook 2017: From Poverty to Prosperity. World Energy Outlook Special Report.

Jamasb, T., R. Nepal, and G. R. Timilsina, 2017. A quarter century effort yet to come of age: a survey of electricity sector reform in developing countries. Energy Journal 38(3): 195–234.

Jones, C. I., and P. J. Klenow. 2016. Beyond GDP? Welfare across countries and time. American Economic Review 106(9): 2426–2457.

Kander, A., P. Malanima, and P. Warde, 2014. Power to the People: Energy in Europe over the Last Five Centuries. Princeton University Press.

Lee, K., E. Miguel, and C. Wolfram, 2020. Does household electrification supercharge economic development? Journal of Economic Perspectives 34(1): 122–144.

Maddison, A., 2001. The World Economy: A Millennial Perspective. Paris: OECD.

Madlener, R., 2009. The economics of energy in developing countries. In: L. C. Hunt and J. Evans (eds.), International Handbook on the Economics of Energy, Edward Elgar.

McCulloch, N., and D. Zileviciute, 2017. Is electricity supply a binding constraint to economic growth in developing countries? EEG State-of-Knowledge Paper Series 1.3.

Solow, R. M., 1956. A contribution to the theory of economic growth. Quarterly Journal of Economics 70: 65–94.

Stern, D. I., 1997. Limits to substitution and irreversibility in production and consumption: a neoclassical interpretation of ecological economics. Ecological Economics 21: 197–215.

Stern, D. I., 2011. The role of energy in economic growth. Annals of the New York Academy of Sciences 1219: 26–51.

Stern, D. I., and A. Kander, 2012. The role of energy in the industrial revolution and modern economic growth. Energy Journal 33(3): 125–152.

Toman, M. A., and B. Jemelkova, 2003. Energy and economic development: An assessment of the state of knowledge. Energy Journal 24(4): 93–112.

van Benthem, A. A., 2015. Energy leapfrogging. Journal of the Association of Environmental and Resource Economists 2(1): 93–132.

Wrigley, E. A., 1988. Continuity, Chance, and Change: The Character of the Industrial Revolution in England. Cambridge: Cambridge University Press.

Data for "Interfuel Substitution: A Meta-Analysis"

I've long thought that there was an error in the way I calculated the shadow elasticity of substitution (SES) in my 2012 paper on interfuel substitution in the Journal of Economic Surveys. This would have been a big problem as the paper carries out a meta-analysis of SESs. But no primary paper reported the results in terms of the SES. I computed all this data from the various ways results were presented in the original studies. I never got around to doing anything about it or even checking carefully whether there was a mistake. I suppose this is because I hate finding mistakes in my papers and as a result procrastination goes into superdrive.

Yesterday a student wrote to me and requested the data. I have now checked the derivation of the SES in my database and also computed it in an alternative way. There is in fact no mistake. This is great news!

The reason I thought that there was a mistake is because of the confusing notation used for the Morishima Elasticity of Substitution (MES). Conventionally, the MES is written as MES_ij for the elasticity of substitution between inputs i and j when the price of i changes. By contrast, the cross-price elasticity is written eta_ij for the elasticity of demand for the quantity of input i with respect to the price of input j!*

I have now uploaded the database used for the meta-analysis to my data website. The following is a description of what is in the Excel spreadsheet:

Each line in the main "data" worksheet is for a specific sample/model in a specific paper. Each of these typically has multiple elasticity estimates.

Column A: Identification number for each paper.

Columns B to L: Characteristics of the authors. Including their rank in the Coupe ranking that was popular at the time.

Column M: Year paper was published.

Columns N to V: Characteristics of the journals in which the papers were published. This includes in Column O the estimated impact factor in the year of publication. Others are impact factors in later years.

Column W: Number of citations the paper had received in the Web of Science at the time the database was compiled.

Column X: Number of citations the lead author has had in their career apart from for this paper.

Columns Y to AO: Characteristics of the sample used for the estimates on that line. So looking at the first line in the table, as an example, we have:

Data from Canada for 1959-1973. Annual observations. This is a panel for different industries. N=2, so there are two industries but a single estimate for both. T is the length of the time series dimension. Sample size is N*T*Number of equations - i.e if there are 4 fuels usually 3 equations are estimated. This could be different if the cost function itself is also estimated, but it looks like no papers did that. (There are also papers using time series for individual industries etc and cross-sections at one point in time.)

Column AH: Whether fixed effects estimation was used or not (only makes sense for panel data).

Column AC: The standard deviation of change in the real oil price in that period.

Column AD: PPP GDP per capita of the country from the Penn World Table. Probably the mean for the sample period.

Column AE: Population of the country in millions. Looks like the mean for the sample period.

Columns AP to AZ are the specification of the model:

Column AP: Not4 - if there weren't 4 fuels in the analysis.

Column AQ: Partial elasticity - this is holding the level of total energy use constant.

Column AR: Total elasticity - this allows the level of total energy use to change.

Columns AS and AT: If this is a dynamic model these are estimates of the short-run or the long-run elasticity.

Column AU: The model is derived from a cost function, or something else.

Column AW: Functional form of the model.

Column AW: Form of the equations estimated - usually cost shares - log ratios means the log of the ratio of cost shares.

Column AX to AZ: How technical change is modeled. Many papers don't model any technical change explicitly. Energy model means there is biased technical change for energy inputs. Aggregate model means that if other inputs are also modeled they also have biased technical change. Kalman means that the Kalman filter was used to estimate stochastic technical change.

Columns BA to the end have the actual estimates. Different papers provide different information. All the various estimates eventually are converted into Shadow Elasticities of Substitution. 

Columns BA to BP: Own price and cross-price elasticities of demand. For example: Coal-Oil means the cross-price elasticity of demand for coal with respect to the price of oil.

Columns BQ to CF: Reported translog cost function parameters.

Columns CP to CS: Cost shares at the sample mean. These are used in various elasticity formulae. They were derived in a variety of ways from the information in papers. One of these methods is the quadratic solution in Columns CG to CO. It uses demand elasticities and translog parameters to reverse engineer the cost shares. Other estimates take the ratio of demand and Allen elasticities.

Columns CT to DE: Morishima elasticities of substitution. These are asymmetric - so we have oil-coal and coal-oil. Here the terminology is very confusing. The standard terminology is that MES_ij is for a change in the price of i. So coal-oil is for a change in the price of coal. This is the reverse of what is used for cross-price elasticities! It is super-confusing.

Columns DF to DK have the shadow elasticities I actually used in the meta-analysis.

Columns DL to EA have the Allen elasticities of substitution. Some of these are reported in the papers and some I computed from the cross-price elasticities.

* You can learn more about all these elasticities in my 2011 Journal of Productivity paper on the topic.

Do Energy Efficiency Improvements Reduce Energy Use? Empirical Evidence on the Economy-Wide Rebound Effect in Europe and the United States

We have just posted a new working paper on RePEc and SSRN extending our structural vector autoregression methodology for estimating the economy-wide rebound effect and applying it to several European countries as well as the United States. I coauthored the paper with Anne Berner at University of Göttingen, Stephan Bruns at Hasselt University, and Alessio Moneta at the Sant'Anna School of Advanced Studies in Pisa. 

We developd this approach as part of our DP16 Australian Research Council funded project on energy efficiency. This is a multivariate time series model using time series for energy use, GDP, and the price of energy. The model allows us to control for shocks to GDP and the price of energy but to model the responses of those variables to the energy efficiency shock. 

We estimate the effect of an energy efficiency shock on the use of energy. Initially, energy use falls, but we found using U.S. data that it then ends up bouncing back to almost where it started. This means that the rebound effect is around 100%. Energy efficiency improvements don't end up saving energy in the long run. That paper has now been published in Energy Economics.

This new paper extends this research in two ways:

1. We control for a wide array of macroeconomic variables that might affect our key variables of interest. In order to squeeze all that information into our model, we carry out a factor analysis and use the first two principal components. This time series model incorporating these factors is called a Structural Factor-Augmented Vector Autoregressive (S-FAVAR) model. The extracted principal components for our five countries are shown in this figure:

2. We apply the model to five countries rather than just the United States. The downside is that we ended up with much shorter time series, only covering 2008-2019.

We also use a Kalman filter method to derive monthly GDP series for the European countries. The choice of countries was restricted by the availability of reliable energy data. As we didn't have separate monthly primary electricity data for the European countries, our energy variable for these countries is just fossil fuels.

Our results are quite similar to our previous U.S. study:

The graph on the left shows how energy use changes over time following an energy efficiency shock. In all countries, it bounces back a lot. It seems like there is more chance of permanent energy savings in the UK than in the other countries. On the other hand, in the long run, the 90% confidence interval of the rebound effect overlaps 100% in all countries. So, energy savings aren't large and may be zero in the long run.

Of course, despite including more information, the results depend on a lot of assumptions. Most importantly, we are talking about an improvement in energy efficiency that is uncorrelated with shocks to the GDP such as total factor productivity improvements. It's possible that the rebound to shocks that are correlated to TFP shocks, if they exist, is quite different. Also, energy efficiency policies that get consumers and firms to do costly things to save energy theoretically have negative rebound. They should end up saving even more energy than is mandated. Given our results, these don't seem to be that important, but we shouldn't say that such policies won't save energy.

Energy Efficiency: What Has It Delivered in the Last 40 Years?

I'm one of nineteen authors of a new review of energy efficiency economics. It was commissioned for the Annual Review of Environment and Resources, where it is still in (second-round) review. The team was put together and led by Harry Saunders and Joyashree Roy.

Over the past four decades different disciplinary approaches independently adopted different definitions of energy efficiency to answer specific problems. Even within economics there are at least three different ideas of energy efficiency. Technical efficiency in economics compares the quantity of inputs used to produce given outputs (or vice versa) to the best practice or frontier level. This is a relative measure of energy efficiency. But economists often talk about energy efficiency in absolute terms too,  measured as either simply an increase in energy services per unit input or using the concept of energy augmenting technological change where the amounts of other inputs and the technology associated with them are held constant. Energy augmenting technological change is usually used when modeling economy-wide rebound, whereas the energy services per unit input might be used when investigating the energy efficiency gap.

The energy intensity of economies (a metric measuring energy consumption per unit of GDP), which is often interpreted as a proxy for energy efficiency, has trended downwards (increasing efficiency) globally and in many major economies over the last century. But as panel (a) below shows, in many regions of the world, especially poorer or hotter regions, energy intensity instead increased. Today, energy intensity is more similar around the world than in the past.

Innovation in energy-saving technologies is an important driver in improving aggregate energy efficiency deployment by lowering costs and inducing adoption. The productivity of numerous energy-using products has improved dramatically. (e.g., lighting had a 10,000-fold improvement in lumens/Watt since the start of the industrial revolution). 

Energy efficiency improvements, in themselves, generally increase economic welfare. But when we consider negative externalities, such as pollution emissions, welfare effects are more ambiguous. Interventions such as improperly calibrated subsidies to improve energy efficiency or mandates to use costly technologies can lead to a reduction in household welfare.

There is still uncertainty and difficulty in measuring economy-wide rebound effects. Rebound may limit the ability to reduce or constrain overall energy use. In general, it makes more sense to address the environmental impacts of energy use with specific environmental policies rather than trying to reduce energy use with energy efficiency policies.

The contribution of different factors to the persistent “energy efficiency gap”, i.e., the difference between the energy consumption observed and the potential energy consumption levels that would result from the adoption of cost-efficient energy efficient technologies and strategies, is still not well understood. Market and regulatory failures, departure of consumer behavior from rational choice theory, lack of information, the principal-agent problem, among other issues may all contribute to the energy efficiency gap.

Policy interventions aimed at overcoming or reducing barriers to energy efficiency deployment target behavioral anomalies and perceived market failures. They include provision of feedback to energy users, the use of social norms, commitment devices, rewards and regulatory mechanisms such as taxes, subsidies, building codes, etc. The literature and evidence are mixed on the effectiveness of each of these, but all seem to show promise to some degree. 

Methodological advances for examining energy efficiency effects on energy use have been substantial. Primary advances include randomized control trials coupled with appropriate econometric methods, developments in econometric methods and lab/field experiments, agent-based modeling formulations, general equilibrium methods, and behavioral science. 

The following diagram summarizes the state of knowledge across different scales and the needed scope of future research:

Future research should bring together researchers from different fields to shed new light onto energy efficiency questions. Examples of such endeavors include: (i) at the micro-level, a better understanding of consumer choice and behavior by combining insights from engineering and the advanced metering and sensing infrastructure, with those from micro-economic theory as well as the theory of choice and with behavioral economists’ models; (ii) at the program evaluation level, there is a need to continue to develop methods to understand causal inferences using econometrics as well as machine learning to better understand program outcomes; (iii) at the macro-level, developing flexible and credible general equilibrium models that also capture environmental and climate externalities outcomes, and that have good input data to enable us to understand the dynamics of energy efficiency improvements across the economy, the environment, and society, are needed.

Prepaid Metering and Electricity Consumption in Developing Countries

I've written an Energy Insight policy brief for the EEG Programme with my PhD student Debasish Das on prepaid metering and its effect on electricity consumption.

The bottom line is that consumers who are switched to prepaid metering significantly reduce their electricity consumption. 

Debasish is working on a study of the effect of prepaid metering in Bangladesh and some preliminary results are in this paper. This graph shows the estimated difference in monthly electricity consumption between consumers in two areas of Dhaka, Bangladesh around the time that one group was switched to prepaid metering:

 

 

Electricity consumption in the treated group fell by 17%. This graph didn't make it into the final version of the paper, because it was deemed to be too mathy. Debasish has a very large dataset that he obtained from the Bangladesh electric utilities. He's still working on getting this into a usable form. But hopefully we will have some more results soon.

How Large is the Economy-Wide Rebound Effect?

Last year, I published a blogpost about our research on the economy-wide rebound effect. The post covers the basics of what the rebound effect is and presents our results. We found that energy efficiency improvements do not save energy. In other words, the rebound effect is 100%. This doesn't mean that improving energy efficiency is a bad thing. It's a good thing, because consumers get more energy services as a result. But it probably doesn't help the environment very much.

I now have a new CAMA working paper, which surveys the literature on this question. Contributions to the literature are broadly theoretical or quantitative. Theory provides some guidance on the factors affecting rebound but does not impose much constraint on the range of possible responses. There aren't very many econometric studies. Most quantitative studies are either calculations using previously estimated parameters and variables or simulations.

Theory shows that the more substitutable other inputs are for energy in production the greater the rebound effect. This means that demand for energy services by producers is more elastic and so reducing the unit costs of energy services increases the amount used by more.

The most comprehensive theoretical examination of the question is Derek Lemoine's new paper in the European Economic Review: "General Equilibrium Rebound from Energy Efficiency Innovation." Lemoine provides the first mathematically consistent analysis of general equilibrium rebound, where all prices across the economy can adjust to a change in energy efficiency in a specific production sector. He shows that the elasticity of substitution in consumption plays the same role as the elasticity of substitution in production: the greater the elasticity, the greater the rebound, ceteris paribus.

Beyond that, the predictions of the model depend on parameter values. The most likely case, assuming a weak response labor to changes in the wage rate, is that the general equilibrium effects increase energy use relative to the partial equilibrium direct rebound effect for energy intensive sectors and reduce it for labor intensive sectors.

Lemoine uses his framework and previously estimated elasticities and other parameters to compute the rebound to an economy-wide energy efficiency improvement in the US. The result is 38%. There are two main reasons why the real rebound might be higher than this. First, most of the elasticities of substitution in production that he uses are quite low because of how they were estimated. Second, an energy efficiency improvement in any sector apart from the energy supply sector does not trigger a fall in the price of energy. A fall in the price of energy would boost rebound. This is because there are no fixed inputs and there are constant returns to scale in energy production.

There are similar issues with simulations from computable general equilibrium models (CGE). The assumptions that modellers make and the parameter values they choose make a huge difference to the results. Depending on these choices, any result from super-conservation, where more energy is saved than the energy efficiency improvement alone would save, to backfire, where energy use increases, is possible.

Rausch and Schwerin estimate the rebound using a small general equilibrium model calibrated to US data. This is somewhere between the typical CGE model and econometric models. They use the putty-clay approach to measuring and modeling energy efficiency. Increases in the price of energy relative to capital are 100% translated into improvements in the energy efficiency of new capital equipment. Once capital is installed, energy and capital must be used in fixed proportions. Rebound in this model depends on why the relative price changes. If the price of energy rises, energy use falls. However, if the price of capital falls energy use increases. These are very strong assumptions, which determine how the data are interpreted. Are they realistic? Rausch and Schwerin find that historically rebound has been around 100% in the US.

Historical evidence also hints that the economy-wide rebound effect could be near 100%. Energy intensity in developing countries today isn't lower than it was in the developed countries when they were at the same level of income. This is despite huge gains in energy efficiency in all kinds of technologies from lighting to car engines. This makes sense if consumers have shifted to more energy intensive consumption goods and services over time. Commuters and tourists on trains in the 19th and early 20th centuries have been replaced by commuters and tourists in cars and on planes in the late 20th and early 21st centuries.

I only found three fully empirical econometric analyses. One of them is our own paper. The others are by Adetutu et al. (2016) and Orea et al. (2015). Both use stochastic production frontiers to estimate energy efficiency. This is a potentially promising approach. Adetutu et al. then model the effect of this energy efficiency one energy use, using an autoregressive model. This includes the lagged value of energy use as an explanatory variable, which means that the long-run effect of all variables is greater in absolute value than the short-run effect. As in the short run, energy efficiency reduces energy use, in the long run it reduces it even more. The result is super-conservation even though short-run rebound is 90%. In Orea et al.'s model, the purely stochastic inefficiency term is multiplied by [1-R(γ'z)] where z is a vector of variables including GDP per capita, the price of energy, and average household size. R(γ'z) is then supposed to be an estimate of the rebound effect. But really this is just a reformulation of the inefficiency term – nothing specifically identifies R(γ'z) as the rebound effect.

In conclusion, the economy-wide rebound effect might be near 100%. But I wouldn't describe the evidence as conclusive. Both our research and the historical investigations might be missing some important factor that has moved energy use in a way that makes us think it is due to changes in energy efficiency, and Rausch and Schwerin make very strong assumptions about analysing the data.

Energy Efficiency Improvements Do Not Save Energy

I have a new working paper out, coauthored with Stephan Bruns and Alessio Moneta, titled: "Macroeconomic Time-Series Evidence That Energy Efficiency Improvements Do Not Save Energy". It's another paper from our ARC funded project: "Energy Efficiency Innovation: Diffusion, Policy and the Rebound Effect". We estimate the economy-wide effect on energy use of energy efficiency improvements in the U.S. We find that the rebound is around 100%, implying that in the long run energy efficiency improvements do not save energy or reduce greenhouse gas emissions.

At the micro level, we might naïvely expect a 1% improvement in energy efficiency to reduce energy use by 1%. But people adjust their behavior. Efficiency improvements reduce the cost of energy services like heating, transport, or lighting. Because these are now cheaper to produce, people consume more of them, and so the percentage reduction in energy use is less than the improvement in efficiency. This is known as the direct rebound effect.

People might also redirect their spending to consume more of complementary goods, like larger houses in the case of residential heating improvements, and reduce their consumption of substitute goods and services, like bus rides or cycling, in the case of car fuel economy improvements. These changes have implications for the energy used to produce these goods and services. Additionally, the reduction in demand for energy should lower the price of energy further boosting the rebound in energy use. Finally, the improvement in energy efficiency is an increase in productivity, which should result in economic growth. Higher incomes mean higher demand for energy. Adding these indirect rebound effects to the direct rebound effect we get the economy-wide rebound effect.

The size of the economy-wide rebound effect is crucial for estimating the contribution that energy efficiency improvements can make to reducing energy use and greenhouse gas emissions. Our study provides the first empirical general equilibrium estimate of the economy-wide rebound effect. Previous studies use simulation models, known as computable general equilibrium models, or partial equilibrium econometric models that don't allow the price of energy to adjust. Some of the latter studies also measure rebound incorrectly, for example assuming that energy intensity – energy used per dollar of GDP – measures energy efficiency. In fact, the majority of the rebound effect happens when energy intensity rebounds as people shift to more energy intensive consumption after an energy efficiency improvement. Economic growth induced by the efficiency improvement is expected to contribute less to total rebound.

We use a structural vector autoregressive model, or SVAR, that is estimated using search methods developed in machine learning. We apply the SVAR to U.S. monthly and quarterly data. An SVAR explains changes in the vector of variables, x, in terms of its past values and a vector of serially and mutually uncorrelated shocks, ε:

In our basic model, the vector, x, contains three variables: primary energy use, GDP, and the price of energy. The first of the shocks is a shock to energy use, holding constant shocks to GDP and the price of energy and the past values of all three variables. We think this is a reasonable definition of an energy efficiency shock. The other two shocks are income and price shocks.

The matrix, B, which transmits the shocks to the dependent variables cannot be estimated without imposing some restrictions or conditions on the model. Usually economists use economic theory to impose restrictions on the coefficients in B (short-run restrictions) and the Π_i (long-run restrictions). Alternatively, they sample a range of models, rejecting only those that don't meet qualitative "sign restrictions" on the matrix B. Instead, we use independent component analysis, an approach that is relatively new to econometrics. This imposes conditions on the nature of the shocks instead and estimates B without direct restrictions. Unlike the short- and long-run restrictions approach, it doesn't impose a priori restrictions on the data, and unlike the sign restrictions approach, it estimates a unique model.

Using the estimated SVAR model we compute the impulse response functions of the dependent variables to the shocks:

The top left graph shows the effect of an energy efficiency shock on energy use. The grey shading is a 90% confidence interval, the x-axis is in months, and the y-axis in log units.

Initially, an energy efficiency shock strongly reduces energy use, but this effect wears off over the following years as consumers and the economy adjusts. Eventually, there is no change in energy use so that rebound is 100%.

The other graphs in the first column show the effect of the energy efficiency shock on GDP and the price of energy. The second column shows the effect of a shock to GDP, and the final column an energy price shock.

The implications for policy are that encouraging energy efficiency innovation is unlikely to make a contribution to reducing greenhouse gas emissions. This is one reason why I am skeptical of projections that predict that energy intensity will fall much faster in the future than in the past because of energy efficiency policies.

On the other hand, if these policies raise rather than reduce the costs of producing energy services then the direct rebound (and presumably the economy-wide rebound) will be negative rather than positive. As, apart from their environmental effects, these would reduce economic welfare, it seems that there would be better options to reduce emissions by switching to low carbon energy.