A new report from the US Breakthrough Institute (BTI) provides evidence that historical improvements in the energy efficiency of lighting, steel and electricity production have led to greater energy consumption that would have been the case in the absence of those improvements. In other words, the ‘rebound effects’ have exceeded 100% (‘backfire’). The authors expect this experience to be replicated in industrialising economies, with the result that improved energy efficiency will contribute much less to reducing energy use and carbon emissions than is commonly assumed.

The BTI report was publicised in a New York Times op-ed, leading to various follow-up blogs (e.g. from Joe Romm) and a flurry of activity on Twitter. A similar response greeted at an earlier BTI report on rebound effects which found: “…a large expert consensus and strong evidence that below-cost energy efficiency measures drive a rebound in energy consumption that erodes much and in some cases all of the expected energy savings.” Both reports brought welcome media attention to this important subject, but the intensity of the ensuing debate demonstrates that a consensus on the importance of rebound remains elusive.

‘Rebound effects’ is an umbrella term for a variety of economic mechanisms that reduce the energy and emission savings from improved energy efficiency. At its simplest, improved energy efficiency leads to cheaper energy services, thereby encouraging increased consumption of those services that offsets some of the energy savings. But the story doesn’t end there, as there are also multiple indirect and second-order effects to consider – such as efficiency improvements enabling firms to expand output, lower product prices and increase market demand which in turn stimulates economic growth and aggregate energy consumption.  As BTI observe, rebound effects are the emergent outcome of multiple feedback mechanisms within complex, non-linear economic systems – and the magnitude of those effects may be expected to vary significantly between contexts and over time.

One reason for the continued controversy is that rebound effects are very hard to measure. Experimental studies and econometric analysis of secondary data can isolate a limited number of mechanisms within constrained spatial and temporal boundaries, but estimation becomes more difficult – and causality harder to establish – as those boundaries expand. Energy efficiency improvements are typically associated with broader improvements in product and process technology that make it difficult and perhaps misleading to isolate the impact of energy efficiency improvements alone. The latter can be achieved with energy-economic models that simulate ‘pure’ energy efficiency improvements and track their economy-wide implications, but while these capture a broader range of economic mechanisms their results are sensitive to assumptions with limited empirical foundation.

However, these methodological problems are hardly unique to rebound effects – a host of other physical and economic phenomena present equally difficult challenges. Moreover, our confidence in the nature and likely magnitude of rebound effects is growing rapidly as the volume of evidence expands, new techniques are applied, new data sources are used and previous studies are re-evaluated. Since I reviewed the subject in 2007, the volume of evidence has grown exponentially. An indication of this comes from the IPCC Working Group 3 who, after ignoring rebound effects in their previous four reports, now state that “rebound effects cannot be ignored” and highlights instances where these effects could be large. For example:

“….A comprehensive review of 500 studies suggests that direct rebounds are likely to be over 10% and could be considerably higher …..Other reviews have shown larger ranges with Thomas and Azevedo (2013) suggesting between 0 and 60%. For household‐efficiency measures, the majority of studies show rebounds in developed countries in the region of 20-45% (the sum of direct and indirect rebound effects), meaning that efficiency measures achieve 65-80% of their original purposes …..For private transport, there are some studies that support higher rebounds, with Frondel et al.( 2012) findings rebounds of between 57 and 62%.

There is evidence to support the claim that rebound effects can be higher in developing countries …..Roy (2000) argues that rebound effects in the residential sector in India and other developing countries can be expected to be larger than in developed economies because high‐quality energy use is still small in households in India and demand is very elastic……..However, there is considerable uncertainty of the precise scale of rebound effects in developing countries with more research required……”

Since the studies cited in this quote neglect economy-wide rebound effects, they also provide only a partial picture. Nevertheless, the available evidence does not suggest that rebound effects routinely exceed 100% – and this is not what BTI claims. Indeed, in those sectors where rebound effects have been most extensively investigated – notably car travel in the US – they appear relatively modest (e.g. less than 20%). This conclusion is unsurprising: with saturation levels of car ownership and use in the US, one would not expect improved fuel efficiency to encourage significantly more driving. But this observation rather misses the point. Like the proverbial drunk searching for his keys under the lamp-post because that is where the light is better, searching for rebounds where the data is better can mean failing to look where data is poorer but rebounds are potentially larger – namely within the production side of the economy and within the emerging economies who account for most of the recent and projected growth in global energy consumption.

The BTI report helpfully shifts our attention to these poorly illuminated areas. BTI argue that emerging economies are in an analogous situation to the developed world a century ago – with consumer demand for energy services being far from saturated and with a huge potential for improving industrial energy efficiency – and thereby for increasing productivity, output, wealth and energy consumption. Roger Fouquet has shown how improvements in the efficiency of lighting and passenger transport in the late 19th century led to a more than proportionate increase in energy consumption. BTI expects the same outcome to follow in emerging economies for lighting, electricity generation and steel production. If BTI are correct, this matters a great deal, since these sectors will have far more influence on future global energy demand than anything that happens within developed economies. As Schipper and Grubb observed in 2000: “….the shadow of Jevons lurks here for the same reason that more efficient coal use did not save coal: the combined effects of different rebounds are very important when energy availability, energy efficiency, and energy costs are a significant constraint to activity and therefore energy use…”

Importantly, the potential for the large rebounds is not reflected in the majority of global energy scenarios – including those produced by the Global Energy Assessment and the IEA. As a result, we may be systematically underestimating the already formidable challenge of reducing global carbon emissions. As BTI staff member Alex Trembath points out, if global average rebound effects lie between 20 and 60%, rather than the 9% assumed by the IEA (WEO 2012, p316), their ‘450 scenario’ may underestimate global energy consumption in 2035 by between 3 and 13%. That is equivalent to between 30 and 130% of the total projected contribution from either renewables or nuclear.

For those of us preoccupied by climate change, this is deeply troubling. It raises difficult questions about the plausibility of various mitigation scenarios and of the additional measures that may need to be taken as a result. For BTI, the solution lies in expanding low carbon energy supply, including extensive deployment of nuclear and CCS. Others are more wary of these solutions and, perhaps as a result, more reluctant to acknowledge the possibility of large rebounds. These differences contribute to raising the temperature of the debate. But the BTI report forcefully reminds us that this is a partial way of viewing the issue.

If energy is appropriately priced, cost-effective energy efficiency improvements should be considered desirable, whatever their impacts on aggregate energy consumption. Rebounds are simply economic responses to productivity improvements that raise human welfare. For example, rebounds from household insulation mean that low income households are no longer cold and unhealthy and have more cash to spend on other things. Rebounds from energy efficient lighting mean that poor children in developing countries can read and do their schoolwork in the evenings. Rebounds from fuel efficient stoves mean that low income households can cook a greater variety of foods more often, improving nutrition while still saving money on their food bills. Rebounds within industry mean that firms can cut costs, employ more people and produce more goods, and that more people can benefit from purchasing those goods. Rebounds in electricity generation mean that more people can gain access to lower cost electricity services such as refrigeration, cooling and electronic communications. And so on. This simple message is often lost with the dominant focus upon energy-saving and demand reduction.

So is BTI correct in concluding that “…the high likelihood of backfire in certain sectors in industrialising economies should caution us against heavy reliance upon energy efficiency as a climate mitigation strategy…”? As someone immersed in the practical difficulties of estimating rebound effects, I am more cautious. Although the evidence on rebounds in developing countries is growing, it remains very patchy and there are considerable difficulties in translating the 19^th century experience to modern India or China. However, it seems reasonable to conclude that rebound effects within most sectors of the industrialising world will be significantly larger than those currently seen within developed countries. Since this fact continues to escape the attention of most analysts and policymakers, it also seems reasonable to conclude that meeting global climate targets could be considerably more difficult than we currently assume.

For analysts, there is an urgent need to move away from the ‘well-lit areas’ and to gain a better understanding of where and why rebound effects could be large. For policymakers and energy-efficiency advocates, there is an urgent need to face up to the challenge of rebound effects, and to stop denying or downplaying their importance. This could be an uncomfortable journey for many, as it has been for me. But it has to be pursued.

Steve’s latest paper on rebound effects, co-authored with Mona Chitnis, Angela Druckman and Tim Jackson, is available free from here.