This paper addresses the probable levels of investment needed in new technologies for energy conversion and storage that are essential to address climate change, drawing on past evidence on the rate of cost improvements in energy technologies. A range of energy materials and technologies with lower carbon emissions over their life cycle are being developed, including fuel cells (FCs), hydrogen storage, batteries, supercapacitors, solar energy and nuclear power, and it is probable that most, if not all, of these technologies will be needed to mitigate climate change. High rates of innovation and deployment will be needed to meet targets such as the UK’s goal of reducing its greenhouse gas emissions by 80 per cent by 2050, which will require significant levels of investment. Learning curves observed for reductions in unit costs of energy technologies, such as photovoltaics and FCs, can provide evidence on the probable future levels of investment needed. The paper concludes by making recommendations for policy measures to promote such investment from both the public and private sectors.
In response to increasing scientific evidence of human interference in climate systems, policy makers have started to set challenging long-term targets for reducing emissions of carbon dioxide (CO2) and other greenhouse gases (GHGs). This will require radical changes in systems for delivering services that contribute to human welfare, particularly relating to energy services of lighting, heating and electric power, and transport services of mobility and accessibility. As well as changes to how we consume these services, this will require significant innovation in a range of low carbon technologies and materials. Despite exciting insights arising from research and development in a number of areas, the deployment of these low carbon technologies has been relatively slow and incremental so far.
However, evidence of cost reductions relating to cumulative levels of deployment gives credence to claims that at least some of these alternatives will become competitive with existing, well-developed technologies for meeting these service demands. This paper examines the probable levels of investment needed, by drawing on research investigating these so-called ‘learning’ or ‘experience’ curves. However, recent research has shown that a naive extrapolation of past trends fails to take into account the complexity of the underlying innovation processes. So, while learning curves can be used to provide ‘ball-park’ estimates of investment needs, this research highlights the need for economists and policy-makers to better understand low-carbon innovation systems and processes, in particular by interacting with engineers and entrepreneurs directly involved in the development of these new technologies.
Section 2 of this paper summarizes the current state of the debate on the economic and policy challenges associated with climate change mitigation. Section 3 introduces learning or experience curves, and illustrates their application to solar photovoltaic and fuel cell (FC) technologies. Section 4 puts this work on learning curves into a broader context of innovation systems, drawing out key implications for policy-making. Section 5 reviews estimates of investments needed to decarbonize the global economy. Section 6 concludes by highlighting the lessons for policy-makers.
2. Climate change: economic and policy challenges
Human society is now at a crucial period in international efforts to prevent catastrophic impacts of human-induced climate change. The accumulation of scientific evidence collected and reviewed by the Inter-governmental Panel on Climate Change (IPCC 2007) concludes that the Earth has already warmed by 0.6°C over the last century and that warming of more than 2°C above pre-industrial levels would lead to dangerous impacts on physical, biological and human systems. However, despite agreement at Kyoto in 1997 to reduce industrialized countries’ emissions by 2012, global emissions have continued to rise over the last decade, making it very difficult to achieve anything close to the 2°C target without urgent steps to realize an early peaking and subsequent rapid reductions of global emissions (Anderson & Bows 2008). Nations failed to reach agreement at Copenhagen in December 2009 on a legally binding international treaty on emissions reductions after 2012. However, both industrialized and large emerging economies agreed under the Copenhagen Accord to implement internationally verifiable emissions reductions and mitigation actions, with the aim of meeting the 2°C target (UNFCCC 2009).
The UK has led the way in setting challenging long-term emissions reductions targets, if not yet in implementing some of the hard political choices that follow from these. In 2008, the UK Government, with all-party support, passed the Climate Change Act, which set a goal of reducing the UK’s GHG emissions by 80 per cent by 2050, from 1990 levels. Significantly, the Act also introduced an institutional innovation, the Committee on Climate Change, comprising scientific, technological and economic experts, to advise the Government on five-yearly carbon budgets to put the UK on track to achieve this goal. The Committee’s first report recommended carbon budgets of two strengths for the periods 2008–2012, 2013–2017 and 2018–2022 (CCC 2008). Their interim budget would lead to a 34 per cent reduction in UK GHG emissions by 2020 from a 1990 level (equivalent to a 21% reduction relative to 2005 levels). However, the Committee recommended that, for the UK to be on track to meet its share of a global emissions reduction pathway consistent with the 2°C target, an intended budget was required leading to a 42 per cent reduction in UK GHG emissions by 2020 from a 1990 level (equivalent to a 31% reduction relative to 2005 levels). It argued that it was only realistic for the UK to achieve this intended budget if there was an international agreement at the Copenhagen conference (CCC 2008). As part of the financial budget in April 2009, the Government agreed to implement the interim budget of a 34 per cent reduction in GHG emissions by 2020, to be strengthened if there was a global deal at Copenhagen (HM Treasury 2009). The UK Low Carbon Transition Plan (HM Government 2009), published in July 2009, sets out the measures to achieve this target. These include support for 40 per cent of UK electricity to come from low carbon sources by 2020, including 30 per cent of electricity from renewables, funding of up to four demonstrations of capturing and storing carbon emissions from coal power stations, and facilitating the building of new nuclear power stations, as well as further measures for improving the energy efficiency of homes, businesses and new vehicles.
The recommended UK carbon budgets closely followed the agreement achieved in December 2008 by the European Union member states to achieve a 20 per cent reduction in European GHG emissions by 2020, with a promise of a 30 per cent reduction if there was an agreement at Copenhagen. This was backed by measures to strengthen the carbon emissions caps set to 2020 under the European emissions trading scheme (ETS), as well as to set new targets of 20 per cent of final energy from renewables and a 20 per cent improvement in energy efficiency by 2020 (Official Journal of the EU 2009). Significantly, in the USA, President Obama has called for an 80 per cent reduction in US GHG emissions by 2050 and lent his support to a bill recently introduced in Congress to set up a similar national cap-and-trade scheme in the USA. The Waxman–Markey bill to do this was passed by the US House of Representatives in June 2009, but a similar bill had yet to be passed by the US Senate by the beginning of 2010. Such a bill would provide a strong incentive to the innovation and deployment of a range of low-carbon energy technologies, with potential first-mover economic advantages, provided that other countries followed suit (Porter & van der Linde 1995).
A major impetus to these national and international efforts was provided by the Stern review on the economics of climate change, launched at the Royal Society in November 2006 (Stern 2007). By reviewing both ‘bottom-up’ technology-based and ‘top-down’ macro-economic models, Stern concluded that emissions reductions needed to achieve global mitigation targets could be achieved at an annual cost of 1–2% of global GDP by 2050. He argued that this cost would be much lower than the annual costs and risks of the impacts of climate change, which are estimated to be equivalent to losing 5–20% of global GDP, now and forever, based on integrated assessment modelling (Stern 2007).
Stern argued that accelerating technological innovation is a key component of policies to deliver timely, effective and economically efficient climate change mitigation. Informed by innovation systems thinking (Foxon 2003; Grubb 2004; Neuhoff 2005), Stern identified three key policy areas as all being necessary for accelerating low-carbon innovation:
— carbon pricing, through taxes or tradable permit schemes;
— increasing support for R&D, demonstration projects and early stage commercialization investments of clean technologies; and
— measures to overcome institutional and other non-market barriers to deployment.
As noted, governments are now beginning to implement these types of policy measure, such as the European ETS, and the renewables obligation (in the UK) and feed-in tariffs for support of renewables technologies (in Germany, Spain and many other EU countries; Stenzel & Frenzel 2008). However, policy-makers face the challenge of judging the appropriate level of support for these initiatives and deciding the extent to which they should apply generally or be focused on specific technologies, in the face of financial budgetary pressures and potentially conflicting objectives of ensuring energy security and minimizing increases in energy bills and transport costs (Foxon & Pearson 2007). Hence, reliable information about expected cost reductions in technologies would be very valuable to policy-makers.
3. Learning and experience curves
Learning curves, relating reductions in unit costs of a new technology to increasing efficiency of production, were first recognized in the early aviation manufacturing industry (Wright 1936). Economist Kenneth Arrow argued that this was a result of ‘learning-by-doing’ providing opportunities for cost reductions and quality improvements, derived from the accumulation of experience in production (Arrow 1962a). To also incorporate other aspects of experience, including ‘learning-by-using’ from feedbacks from early use (Rosenberg 1982), learning curves are now sometimes referred to as experience curves, and unit cost reductions are usually plotted against cumulative deployment of the technology. A range of empirical studies have found a power-law relation for these cost reductions, of the following form: and where C is the unit cost of the technology and q represents cumulative deployment. Here b is the learning coefficient, and the learning rate, LR, represents the reduction in unit costs for a doubling of cumulative output (equivalent to 1−PR, the progress ratio). When plotted on a log–log scale, the graph of unit costs against cumulative deployment becomes a straight line.
Analysis of past cost reductions for energy technologies has typically shown empirical learning rates of 10–25%, meaning that a 10–25% reduction in unit costs results from a doubling of cumulative deployment (McDonald & Schrattenholzer 2001). Learning rates are an empirical phenomenon with only a broad theoretical basis in the effects of learning and experience, but recent work by one of the founders of their application to energy technologies has sought to provide a stronger theoretical basis for observed learning rates by quantifying the feedbacks from experience (Wene 2008).
Beginning with the work of the International Energy Agency (IEA 2000), the concept of learning or experience curves has particularly been applied to energy technologies in the early stages of their development, for two main purposes:
— to estimate when these technologies are likely to become competitive with existing sources of energy supply and
— to estimate the levels of investment likely to be needed to achieve such a competitive state.
Of course, such estimates are subject to high levels of uncertainty, as they are highly sensitive to variations in the learning rate and other factors, such as the future level of a carbon price, under a carbon tax or ETS. Nevertheless, they can provide a ‘ball-park’ estimate of the levels of investment needed. This is given by the area under the learning curve, as shown in figure 1 from the Stern review.
(a) Learning curves and investments for solar photovoltaics
Solar photovoltaic (PV) technology—the direct conversion of sunlight into electricity—was one of the first energy technologies for which learning was studied in detail. The IEA found that, for the period 1976–1997, first generation crystalline silicon PV showed reductions in the PV module price as cumulative production increased consistent with a learning rate of 20 per cent (IEA 2000). By 1997, the price of PV modules had reached around $5 per peak watt, Wp, of output, as almost 1 GW of production was reached. By extrapolating this curve, the IEA was able to estimate the levels of investment needed for PV systems to become competitive against centralized power generation using fossil fuels at around $0.5 per Wp. It found that $60 million of learning investments would be needed for PV to achieve this break-even point, when cumulative production of PV would reach around 200 GW.
However, the size of these learning investments was found to be highly sensitive to the projected learning rate. Just a 2 per cent variation on the learning rate either way would result in the size of the learning investment needed varying between $40 billion and $200 billion. It is also important to note that, in this approach, the dependent variable is cumulative production, and so the time at which this break-even point is achieved depends on the level of production achieved. Noting these caveats, the IEA recommended that this approach ‘should be embedded in a continuous process of policy analysis and evaluation, where it will serve as an interactive tool for developing effective strategiesto make environmentally friendly technologies available to the energy system’ (IEA 2000, p. 18).
Since that early work, markets for PV have expanded rapidly, growing at over 40% yr−1 (Maycock 2005). At the same time, R&D on second generation, thin-film PV technologies and third generation, high efficiency PV technologies, has also expanded significantly, and early applications of thin-film technology, such as building-integrated PV, are now emerging. These advances make the extrapolation of PV learning curves subject to even more uncertainty.
Recent work has tried to gain better understanding of the drivers of technical change in PV by disaggregating historic cost reductions into observable technical factors. Nemet (2006) found that the largest contributions to reductions in module costs over the period 1975–2001 came from improvements in module efficiency, increases in production plant size and decreases in the cost of silicon. These improvements in average module efficiency, from 6.3 to 13.5 per cent, resulted from significant R&D investment as well as learning-by-doing; increases in average plant size from 76 kW yr−1 to 14 MW yr−1 were due to economies of scale as well as experience effects; and silicon cost reductions from $300 kg−1 to $25 kg−1 were mainly due to a spillover benefit from the microprocessor industry. These results show that, though unit costs did decrease in logarithmic proportion to increases in cumulative capacity, these cost reductions could not be attributed solely to learning or experience effects. Hence, it has been suggested that experience curve cost estimates should be complemented with bottom-up assessments of the sources of cost reductions (Ferioli et al. 2009). However, a review of cost reduction paths for a number of energy technologies found that the results of the experience curve studies and the bottom-up analysis agree in most cases (Neij 2008), supporting the use of learning or experience curves to estimate future cost reductions and learning investments.
Recent assessments for the level of learning investment needed for PV have produced similar estimates to those of the first IEA report. Schaeffer et al. (2004) found that, extrapolating the historical learning rate of 20 per cent and an annual market growth rate of 20 per cent, would indicate that PV would break even with bulk electricity production by around 2040, requiring learning investments of the order of 600 billion euros. If learning rates of 25 per cent and annual growth rates of 30 per cent could be achieved, then PV could be cost-competitive within 15 to 25 years, with learning investments of around 200 billion euros. However, this would be likely to require higher levels of support for R&D in next generation PV technologies to achieve these learning rates and also support for initial market niches to stimulate market growth, as we shall discuss further below.
(b) Learning curves and investments for fuel cells
FCs enable high efficiency conversion of hydrogen or other fuel source into electricity for either stationary (power or heat and power) or mobile (vehicle propulsion) applications. Different types of FCs are likely to be suitable for different types and scales of application. Polymer electrolyte membrane (PEM) FCs operate at smaller scale (1–25 kW) and low temperatures, and so can be used for either vehicle propulsion or small-scale generation. Molten carbonate, solid oxide and phosphoric acid FCs (MCFC, SOFC, PAFC) operate at larger scale (up to a few MW) and higher temperature, and so are used for stationary combined heat and power (CHP) generation. However, like PV, FCs currently have high costs relative to conventional, fossil fuel based technologies, and so there is much interest in cost reduction potentials. The cost of installing FCs for power generation currently ranges from approximately 4000 to 14 000 euros per kW (Krewitt & Schmid 2004). For example, 200–300 kW stationary MCFC or SOFC systems cost between $12 000 and $15 000 per kW (50% for stack; IEA 2007). Learning rates of 15–25% have been estimated, but as the levels of deployment are currently very small, there is little empirical data by which to assess these (Neij 2008).
For automotive application, Tsuchiya & Kobayashi (2004) examined the potential for cost reduction of a PEM FC stack from the current cost of just under $2000 per kW. They found that, with an estimated learning rate of 21 per cent, a global market penetration of 5 million FC vehicles by 2020 would reduce the FC stack cost to below $50 per kW, in order to make them cost competitive with petrol or diesel internal combustion engine vehicles. Their analysis of cost structures found that the bipolar plates and electrodes contributed the largest share of stack costs, but that the share of platinum costs increases up to 11 per cent of the total costs. A similar study by the IEA (2005) found that to reduce the costs of PEM FCs to $50–100 per kW would require
— the mass-production of membranes and possibly the use of new materials (other than Nafion);
— the mass-production of electrodes based on the new gas diffusion layer technology;
— the mass-production of either plastic or coated-steel bipolar plates;
— achieving an increase in power density from 2 to 3 kW m−2; and
— the production of 100 000 m2 yr−1 of FC stacks, which equals 4000 vehicles per year for an 80 kW vehicle.
The IEA study estimated the learning rates for a FC vehicle of 15 per cent to 22 per cent, including power electronics and other balance of plant costs. The cumulative incremental costs of a global roll-out of FC vehicles to reach 7 million vehicles by 2050 would then be in the range of $1–2.3 trillion (IEA 2005). Again, both R&D to reduce the costs and improve efficiency of the PEM FC, and niche market application to gain experience and reduce the hydrogen fuel system and infrastructure costs would be needed to achieve this.
4. Learning curves in an innovation system context
As we have seen, learning curves illustrate the empirical observation that costs of a new technology decline with the deployment of that technology. This leads us to confidently expect that low-carbon technological alternatives, which are at the very early stages of their potential deployment, will significantly reduce in cost as they become taken up in commercial market contexts. Using learning curves in this way enables us to make ball-park estimates of the levels of investment needed for these technologies to become commercially competitive.
However, this research also shows that we should be careful in drawing too naive conclusions of the ease of achieving widespread deployment of these technologies. Many low-carbon options are in fact families of technologies, which may be at different stages of development and may have different scales or forms of application. Some learning may be shared across these different families, while some may be more specific to particular types or applications. These findings imply that it will be necessary to combine learning curve analysis with more bottom-up component level analysis, and with more systemic analysis of the drivers and barriers to innovation from an innovation systems perspective. By incorporating these insights, learning curves can still play a valuable role in summarizing and communicating insights from more complex perspectives.
A range of research looking at technological learning within an innovation systems perspective (Foxon et al. 2008; Grubb et al. 2008; UKERC 2009) agrees on a number of findings important for policy-making, which are addressed in more detail below:
— it is important to support both R&D and early stage deployment of technologies, as both learning-by-research and learning-by-doing are likely to be important to achieving cost reductions;
— a significant carbon price, through a tax or trading scheme, is a necessary but not sufficient condition for achieving widespread deployment;
— support for niche markets to facilitate early stage commercialization is likely to be crucial for learning and cost reductions to occur; and
— incumbent fossil fuel technologies are also likely to experience cost reductions, through ‘sailing ship’ effects.
(a) Supporting both R&D and learning-by-doing
Though learning curve analysis shows the reduction of unit costs with deployment, more detailed analyses of technological cost reductions also emphasize the importance of R&D (Sagar & van der Zwaan 2006; Jamasb 2007; Jamasb & Köhler 2008). The case for public support of R&D is well known—private firms are unlikely to be able to capture the full benefits of their investments, as new knowledge may be relatively easy to copy once produced, and so firms will invest less in R&D than would be socially optimal (Arrow 1962b). However, public energy R&D in the UK and other IEA countries declined rapidly in the 1980s and 1990s as formerly publicly owned energy companies were privatized. This coincided in the 1990s with significant levels of investment in gas-fired electricity generation leading to high levels of capacity in the UK, reducing incentives for private sector R&D. Levels of both public and private sector R&D investment have begun to rise more recently, stimulated by both awareness of the need for carbon emissions mitigation and a potential generation gap in the UK by around 2016, as most of the existing nuclear generation capacity reaches the end of its life and many coal-fired power stations have to close to meet European standards for emissions of local pollutants. For example, the UK Energy Technologies Institute will provide support to accelerate the pace and volume of research directed towards the eventual deployment of promising low carbon technologies, through matched public and private funding of up to £100 million per year for 10 years. Nevertheless, based on OECD/IEA data, the Stern review recommended a doubling of public energy R&D across IEA countries to $20 billion per year, spread across a wide range of low-carbon technologies (Stern 2007, p. 423).
(b) Carbon price necessary but not sufficient
Until recently, emissions of CO2 and other GHGs have been unpriced, or an ‘externality’ in economic jargon, meaning that there has been no economic incentive to reduce emissions. In response to the challenge of climate change, governments are now putting in place mechanisms to price carbon emissions. Though some economists favour a direct carbon tax on emissions as the simplest mechanism to do this, the European Union favours a carbon ETS. Under such a scheme, an overall cap on emissions is agreed and tradable allowances are distributed or auctioned to participants, who have to buy additional allowances from other participants if they exceed their allocation of the overall cap. In this way, the carbon price is determined by the supply of allowances (dependent on the overall cap) and the demand for allowances (dependent on how easily participants can make their allocated emissions reductions). In relation to investments in new technologies, as figure 1 shows, a higher carbon price makes established technologies more expensive and so reduces the learning investment needed to make the new technology cost competitive. However, as figure 1 also illustrates, to stimulate investment in new low-carbon technologies on its own, a very high carbon price would be needed, which would likely be politically and socially unacceptable. Hence, additional direct support for deployment will be needed. The Stern review estimates that global support for deployment of renewables, biofuels and nuclear energy will have to increase two- to fivefold over the next 20 years from the current level of $33 billion per year (Stern 2007, p. 422).
(c) Support for niche markets
Early views of innovation relied on a linear model, in which ‘technology push’ from R&D met up with ‘demand pull’ from markets, to give rise to a flow from invention and first innovation to diffusion. It is now recognized that innovation is a more complex, dynamic and systemic process, involving multiple actors and interactions within an innovation system (Foxon 2003). This view recognizes that existing high-carbon technologies have accrued economic and social advantages thanks to many years of positive scale, learning, adaptation and network effects, leading to cost reductions, development of complementary institutional rules and frameworks, and individuals’ habits and firms’ routines being aligned to expectations of the continued dominance of these technologies—referred to as ‘carbon lock-in’ (Unruh 2000).
Research examining past transitions away from other similarly dominant technological regimes has identified the importance of niche markets, where both technological learning and social learning can occur (Geels 2002). In addition to performance improvements and price reductions from learning-by-doing and learning-by-using, this could help to align habit, routines and institutional rules with the new technology. For example, distributed forms of electricity generation, such as PV and FCs, may require more active involvement of end-users with maintaining the performance of the technology and also appropriate pricing structures for the sale of excess electricity generated back to the grid (Nye et al. 2010). Unfortunately, UK government support for households and businesses to install microgeneration has so far failed to create strong niche markets because of the stop–start nature of support programmes. For example, in March 2008, the UK government announced that it was suspending grant applications for support for installation of PV to public sector buildings under phase 2 of its low carbon buildings programme, after it had spent PV’s share of the £131 million fund. The introduction of ‘feed-in tariffs’ to provide payments for small-scale renewable electricity generation (from April 2010) and renewable heat production (from April 2011) aims to stimulate more rapid take-up of these technologies by providing a higher and more secure return on the initial investment (DECC 2010).
Niche markets for FCs have been identified as domestic scale micro-CHP for stationary applications and buses and delivery vans in urban settings for vehicle applications (IEA 2005). An innovation system perspective also helps to identify the complementary systemic changes needed for successful growth and deployment of new innovations (Jacobsson & Bergek 2004). As well as creating niche markets, this includes institutional changes to facilitate entry of firms into markets, developing a skills base and a favourable R&D and industrial policy environment, with a key role played by lobbying from ‘advocacy coalitions’ of early-mover private firms and public bodies. Research shows that these conditions for the take-off of FCs in residential CHP markets differ significantly between different countries, with Japan and Germany having the most favourable combination of early markets, public support and private advocacy groups (Brown et al. 2007). On the other hand, UK innovation systems for a range of renewable and low-carbon energy technologies, including FCs, have exhibited ‘systems failures’ in bringing these technologies through from R&D to commercialization (Foxon et al. 2005).
(d) Cost reductions by incumbents
The theory of learning curves relates cost reductions to learning and experience gained through deployment, and so we would expect that incumbent technologies would largely have already realized these gains. However, historical analysis shows that the advent of a competing new technology may often stimulate renewed innovation in an incumbent technology. This is generally known as the ‘sailing ship effect’, after it was first noted in the case of the technological and organizational innovations in the design and use of sailing ships, such as new hulls and labour-saving machinery for rigging, in response to competition from steamships (Gilfillan 1935; Geels 2002). This phenomenon was found by McVeigh et al. (1999) to help explain the relatively low market shares obtained by renewable energy technologies, including biomass, geothermal, solar PV, solar thermal, and wind, in the 1990s. They found that, though the costs of these technologies had fallen in line with, or exceeding, prior projections, the costs of incumbent fossil fuel-based technologies had also fallen, slowing down the deployment of renewable technologies. This was due to both declining technical costs of production and regulatory and institutional changes that expanded competition and reduced costs.
This analysis shows the importance of considering the whole energy system when assessing whether the cost reductions in new low-carbon energy technologies due to learning and experience effects are likely to lead to high levels of deployment. The increasing volatility and high peaking of oil and gas prices in 2008 could indicate that further cost reductions in the costs of fossil fuel technologies may be becoming harder to achieve. Nevertheless, the sailing ship effect shows that incumbent technologies can be expected to respond to the challenge from renewable energy technologies, for example, through investment in demonstrating the potential of capturing and storing carbon emissions from coal- and gas-fired electricity generation.
5. Investment for a low carbon economy
In its World Energy Outlook 2008, the IEA estimated that to meet growing energy demands from rapidly industrializing countries while tackling climate change would require $26.3 trillion of investment globally to 2030, or over $1 trillion per year (IEA 2008). Bottom-up energy economic analysis done by Anderson (2006) for the Stern review provided an estimate of the total learning investment needed to decarbonize the global energy system, assuming that costs of a range of technologies continue to reduce along their learning curves with cumulative deployment. Anderson found that global carbon emissions from energy supply could be reduced from 24 GtCO2e per year in 2002 to 18 GtCO2e per year in 2050 at an annual cost rising from $134 billion in 2015 to $930 billion in 2050. Assuming that the global economy continues to grow at an average growth rate of 2.5 per cent over this period, this would be equivalent to around 1 per cent of global GDP in 2050 (Stern 2007, p. 260). Given slightly more optimistic assumptions about the rate of cost reductions of key technologies, this figure could be reduced to give a net positive benefit to the global economy, thanks to the stimulus to the wider economy given by this level of investment in mitigation technologies (Barker et al. 2008). Macroeconomic models incorporating technological learning produce similar estimates of annual costs of 1–2% of global GDP by 2050 to reduce global GHG emissions by 50 per cent, in order to put the world on track to stabilize atmospheric concentrations between 450 and 550 ppm CO2e (Stern 2007).
The current economic crisis makes achieving these levels of investment harder in the short term (Bowen et al. 2009a). As well as directly reducing the availability of capital for investment, the economic recession reduces demand and emissions, making carbon caps easier to achieve, so reducing the carbon price under the EU ETS. This reinforces the need for early stage deployment support to complement the incentive to invest in low-carbon technologies provided by the carbon price. The Committee on Climate Change noted that the significant emissions reductions produced by the recession could lead to an over rosy impression of progress towards meeting shorter term targets, and that a step change in the rate of emissions reductions is needed to ensure that the UK remains on track to meet its long-term targets (CCC 2009). Stern and others have argued that support for low-carbon technologies should form a larger part of fiscal stimulus packages in the UK and other countries, following the lead of the Obama administration in the USA (Bowen et al. 2009b; UNEP 2009).
As noted, given the large range of uncertainties and the need for choices to be publicly acceptable, it is not justifiable to use analysis of investments based on learning curves to come to conclusions about the exact mix of technologies that should be deployed. Nevertheless, it is notable that, as with other similar analyses, Anderson (2006) found that the largest contribution to carbon abatement would come from improvements in the efficiency with which energy is supplied and used. In addition to large contributions from carbon capture and storage, nuclear power and biofuels, this analysis found significant abatement ‘wedges’ (Pacala & Socolow 2004) coming from solar PV and decentralized CHP, including FCs. Thus, while avoiding trying to pick particular technological winners, analysis based on learning curves reinforces arguments for a step change in public support for both R&D and niche market learning for PV and FCs, as options among a range of low carbon technologies that will be necessary for climate change mitigation.
This paper has shown that high levels of both public and private investment will be needed in new low-carbon technologies for energy conversion and storage. These levels of investment should give rise to cost reductions that will lead to these alternatives becoming competitive with existing technologies over time. Evidence has shown that learning and experience effects, together with economies of scale, are likely to lead to sharp reductions in the costs of these technologies as they are commercially deployed, with learning rates typically showing up to a 20 per cent reduction in unit costs with a doubling of deployment. However, research on learning in an innovation system context shows that public support for both R&D and early stage deployment will be needed, that putting a price on carbon emissions through taxes or trading schemes will be necessary but not sufficient for achieving this, and that support for niche markets will be crucial. This research also shows that incumbent fossil fuel technologies are likely to experience cost reductions, through ‘sailing ship’ effects, as they respond to the challenge of competing new low-carbon technologies.
Estimates suggest that annual costs of 1–2% of GDP would enable energy investment to be targeted towards the development and deployment of low-carbon technological alternatives, in order to meet challenging national and global targets for mitigating human-induced climate change. It is important for this level of investment to be maintained, despite the current economic difficulties, as the analysis by the Stern review and others argues that the costs and risks of the impacts of climate change would be much higher than this. Though the Copenhagen climate change convention failed to produce a legally binding international agreement, the Copenhagen Accord at least signalled a willingness by the world’s largest emitters to make significant emissions reductions. However, a step change in the level of investment in low-carbon technologies by national governments, alongside measures to set a carbon price and demonstration of the potential economic benefits to first movers to a low-carbon economy, will be needed to achieve the internationally agreed target of limiting global temperature rise to 2°C.
The author would like to thank Richard Catlow, Peter Bruce and Peter Edwards for their kind invitation to speak at the Royal Society Discussion Meeting on ‘Energy materials to combat climate change’ on 8–9 June 2009, on which this paper is based. He would also like to thank the organizers and other participants of that meeting for a stimulating event, and two anonymous reviewers for helpful comments on the first draft of this paper.
One contribution of 13 to a Discussion Meeting Issue ‘Energy materials to combat climate change’.
- © 2010 The Royal Society