Energy efficiency improvement will not dramatically change global energy use and related emissions. However, it might promote economic growth.
Anthropogenic greenhouse gas emissions have driven the historical climate change and will continue to drive the future climate change.
Many regions, including the European Union, have put their faith in energy efficiency improvement, trusting it will reduce emissions of carbon dioxide (CO2) from fossil fuel combustion. Typically, a ten per cent improvement in energy efficiency by consumers is expected to reduce energy use and related CO2 emissions by ten per cent in prevailing energy use forecasts.
Reality, however, performs differently.
According to a new study (Wei and Liu, 2017), a ten per cent improvement in energy efficiency for all final energy use at the global level may lead to an “actual” reduction in energy use and related emissions, in the long term, as low as three and one per cent, respectively.
The “take-back” or rebound phenomenon at work here implies that the energy efficiency improvement itself will unlikely effectively change global energy use and related emissions, although it can promote economic growth.
This counter-intuitive rebound phenomenon occurs due to a series of behavior changes of final energy consumers (Figure 1). An energy efficiency improvement reduces the “effective” price of given energy services required by consumers and directly leads to reduced demand for energy goods. How? Consumers achieve a double benefit: first, from cheaper energy services and second, from their saved expenditures for the same energy services as before.
Consumers spend their money elsewhere
Cheaper energy services may stimulate consumers to increase their consumption of energy goods compared to other goods, while energy expenditure savings can be used to increase consumption of both energy and other goods. These behavior adjustments further lead to price changes in energy and other goods markets.
As a result, the whole economy responds to the price changes by adjusting production activities, reallocating resources across sectors and regions, creating innovative activities to adapt to the energy efficiency improvement, and increasing effective labor and capital supply to fully utilize the potential energy services associated with the energy efficiency improvement.
Consequently, the final energy use and related carbon emissions deviate from the original situation. At one extreme, consumers may reduce energy consumption by more than the energy efficiency improvement, leading to super-conservation. Alternatively, consumers may consume more energy than the stipulated energy efficiency improvements, leading to an energy efficiency backfire. Most probably, there will be a situation between these two extremes.
Zooming in on the results
Economists can examine the rebound phenomenon on various scopes and levels. Direct and indirect rebound effects correspond to changes in sectoral (particularly household) energy consumption assuming no effect on other sectors and market prices. Macroeconomic rebound effect refers to regional/global “take-back” effects caused by inter-sectoral links and changes in market prices. Economy-wide rebound effect refers to the sum of direct, indirect, and macroeconomic rebound effects.
Together with my colleague Yang Liu, I have recently examined the size of the global economy-wide rebound effect in the long term using the Global Responses to Anthropogenic Change in the Environment (GRACE) Model. GRACE is a global, recursive dynamic computable general equilibrium (CGE) model that divides the world into eight regions: United States, European Union, Japan, Russia, China, India, Brazil and the rest of the world.
Simulation results indicate that there is a considerable long-term rebound effect in energy, ranging from 55 per cent in Brazil to close to 80 per cent in India, with the global effect close to 70 per cent by 2040. Although at first glance such estimates may seem high, the results are broadly in line with previous estimates.
In 2009, Terry Barker and his colleagues, for example, reported the global rebound effect in energy to be 31 per cent by 2020 and 52 per cent by 2030. The rebound effects in terms of emissions are significantly higher at between 75 per cent (Russia) to 98 per cent (China) and the global estimate is at 90 per cent.
When we introduce energy efficiency as a parallel increase for all final energy users in the model, fossil fuels – cheap compared to renewable energy sources – will only become cheaper, thereby inducing increased consumption of emission-intensive sources.
Our results also illustrate that labour mobility contributes positively to the rebound effect. In fact, when the labour supply is fixed, the global rebound effect on energy falls from 70 per cent to 60 per cent. Interestingly, the rebound effect dropped significantly when the labour supply was fixed in developed countries and China, while experiencing little change in the case of India. This may imply that the labour supply fulfils different roles across regions.
Energy efficiency improvement in the demand side can serve as an effective policy to promote economic growth, but probably cannot itself be an effective policy to reduce the global energy use and related emissions.
In the long term, energy efficiency improvement does not mainly reduce energy use, but instead promotes considerably economic growth and social welfare through inducing additional supply of other productive resources such as labour and capital. To make energy efficiency improvement an effective policy for reduction in energy use and related emissions, policy to improve efficiency of renewable production and consumption alone could be effective for reduction in fossil fuel use and related emissions, but may still not be effective for reduction in total energy use.
Several steps could improve the estimation of global rebound effect. It might be more relevant to policy making if investment costs to obtain the efficiency improvement are considered in the analysis. Importantly, our results depend on various assumptions on substitution elasticities and other parameters, not based on historical data.
Hence, we could validate our simulation model with historical data, and later use them to examine historical rebound effects and re-estimate the long-term global rebound.