Induced technical change: Evidence and implications for energy-environmental modelling and policy†
Working Paper Department of Applied Economics Cambridge University
*
Professor Michael Grubb T.H.Huxley School, Imperial College, London,1 and Senior Research Associate, Department of Applied Economics, University of Cambridge 2
Dr Jonathan Koehler Department of Applied Economics University of Cambridge2
With additional contributions from: Professor Dennis Anderson TH Huxley School Imperial College, London1
†
This report is based on a report to the Deputy Secretary-General, Organisation for Economic Cooperation and Development, Paris, and funding from the OECD is gratefully acknowledged. 1
Imperial College of Science, Technology and Medicine, T.H.Huxley School, Royal School of Mines, Prince Consort Rd, London SW7 2BP.
[email protected] 2 Department of Applied Economics, Sedgewick Avenue, Cambridge
2
Contents Executive Summary.................................................................................................................... 4 I. Introduction............................................................................................................................. 6 2. The importance of technical change ........................................................................................ 7 3. Sources of technical change: classifications and evidence ....................................................... 9 4. Technology modelling in major economy / energy / IA models ..............................................12 5. Explicit studies of induced technical change ..........................................................................16 5.1. Modelling of learning-by-doing: implications for costs, targets and scenarios..................17 5.2. Modelling of investment in knowledge/innovation ..........................................................21 5.3. Insights from simplified parameter models ......................................................................22 6. Incorporation of induced technical change in major macroeconomic, energy sector and IA models.......................................................................................................................................24 6.1 Macroeconomic E3 Models..............................................................................................24 6.2 Energy Sector Models......................................................................................................25 6.3 Integrated assessment Models (IAMs)..............................................................................26 6.4 Macroeconomic Models with Endogenous Technical Change...........................................26 6.5 Shortcomings of the current approaches and directions for further research ......................27 7. Overview of the policy and research implications of induced technical change.......................29 7.1 Aggregate cost implications .............................................................................................29 7.2 Timing and the cost of delay ............................................................................................29 7.3 First mover economics .....................................................................................................30 7.4 Policy instruments and cost distribution ...........................................................................30 8. Conclusions...........................................................................................................................33 Appendix A. Mathematical approaches to modelling endogenous technical change....................35 References.................................................................................................................................39
3
Summary Technical change is important in addressing environmental issues, and becomes particularly so concerning difficult, long-term and global issues like climate change. This conclusion is confirmed by data on specific technologies identified, and by and almost all economic modeling sensitivity studies. However, the discussion often obscures a key distinction between: • autonomous technical change, which depends mostly on autonomous trends and government R&D; and • induced technical change, which depends mostly upon corporate investment (R&D, and learning-by-doing) in response to market conditions. There is very strong evidence that much technical change in the energy sector is of the second kind. Evidence comes from studies of specific technology costs as a function of market conditions and investment; the generation and persistence of energy efficiency potentials including international comparisons; and energy demand trends through the price shocks. Action to limit emissions or otherwise create markets for cleaner technologies can therefore be expected to result in technological improvements including cost reductions. Incorporating such induced technical change (ITC) in economic models (making it endogenous to the models) is very complex. It makes the modeling inherently non-linear and complex, with path dependencies and the potential for multiple equilibria. Such treatment is beyond the scope of the major E3 models currently in use. In these models, technical change is incorporated through exogenous assumptions and this is potentially an important weakness. Specific models devoted to investigating ITC have emerged during the mid and late 1990s. These models show that it can alter results in many ways. Models that draw upon the empirical engineering literature tend to show very large impacts indeed. Models in which innovation is a constrained resource that may be shifted from one sector to another tend to show lesser effects. Other and simpler approaches also show results that appear quantitatively quite divergent. These differences require further exploration but do not undermine the conclusion that induced technical change is likely to be important in the context of long-term issues like climate change. New efforts are seeking to incorporate ITC in more mainstream E3 models but results are not yet available that enable the impact of ITC to be estimated in these models. Overall the economic and policy implications of ITC appear to be as follows; the findings are summarised in Table E1 (drawn from the concluding section): Long run costs. Several (but not all) studies incorporating ITC suggest that it could make addressing climate change – including atmospheric stabilisation - quite cheap in the long run. Within the scope of estimated parameter uncertainties, the innovation could dominate and net economic impacts (irrespective of climate damage) can be positive in some (a minority) of cases. Policy instruments and cost distribution. Far more attention should be given to technical change, but ITC greatly broadens the scope of technology-related policies. Efficient responses may involve a wide mix of instruments, targeted to spur market-based innovation in relevant sectors, and broader mitigation policies including economic instruments. It may not be optimal to equalise marginal costs in each period because the returns to learning-by-doing will differ between sectors and technologies.
4
Table E1. Comparison of implications of autonomous vs. induced technical change Autonomous technical Induced technical change change Process:
Modeling implications: Modeling term Typical main parameter Mathematical implications Optimisation implications
Economic / policy implications: Implications for long-run economics of large-scale problems (e.g. climate change) Policy instruments and cost distribution
Timing implications ‘First mover’economics Spillover / leakage implications
Technical change depends mostly on autonomous trends and government R&D
Technical change depends mostly upon corporate investment (R&D, and learning-by-doing) in response to market conditions
Exogenous AEEI / projected costs Usually linear Single optimum with standard techniques
Endogenous Learning rate / progress ratio Non-linear, complex Potential for multiple equilibria, perhaps very diverse, complex techniques
Atmospheric stabilisation likely to be very costly
Atmospheric stabilisation cheaper, may be quite cheap
Efficient instrument is uniform Pigouvian tax + government R&D
Efficient response may involve wide mix of instruments, targeted to reoriented industrial R&D and spur market-based innovation in relevant sectors,. potentially with diverse marginal costs Tendency to accelerate abatement to induce cost reductions Costs with potentially large benefits Positive spillovers may dominate (leakage likely to be negative over time)
Tendency to defer abatement to await cost reductions Costs with little benefits Spillovers generally negative (positive leakage)
Timing. ITC usually increases the benefits of early action, which accelerate development of cheaper technologies. This is the opposite of the result from models with autonomous technical change, which can imply waiting for better technologies to arrive. Numerical studies of CO2 abatement imply that for action taken at present, the benefits associated with ITC may be substantially larger than the direct Pigouvian benefits of CO2 abatement. First mover economics. ITC offers a partial explanation and economic formalisation of the ‘Porter Hypothesis’, that environmental regulation can improve economic competitiveness by stimulating the development of better technologies, as well as institutional and other responses. This may be particularly important in the context of climate change. Spillover and leakage. If climate change mitigation induces improved technologies in the industrialised nations, it is likely that these technologies will diffuse globally. This will result in a positive spillover, that will offset the ‘negative spillover’ usually hypothesised to result from the migration of polluting industries. Empirical data and analytic understanding are still extremely weak in this area. However, preliminary studies suggest that this effect may dominate over time, resulting overall in negative leakage (i.e. reductions in industrialised countries may also result in reduced emissions in the rest of the world), because of the enormous leverage potentially exerted by global technology diffusion over decades.
5
I. Introduction Technical change is generally acknowledged to be an extremely important factor in addressing major environmental issues, particularly large-scale and long-term problems like climate change (IPCC, 1995; Weitzman, 1997). This has been highlighted in work by the OECD, IPCC, and many others. Recent work also highlights how more rapid development of clean energy technology can reduce risks when faced by uncertain long term environmental threats. However, in economic modelling of policy questions in this area, relatively little attention has been paid to how technical change occurs. In many economic models, technical change is incorporated as an exogenous variable: it is reflected through specific assumptions entered as data about improved efficiency and declining costs of certain kinds of technologies through time. This implies a modelling assumption that technical change is mainly anautonomous process: that it happens in ways that do not depend upon other policy or economic variables. In the wider literature on technical change, however it is acknowledged that technical change is not an autonomous process: it occurs as a result of identifiable processes, such as government R&D, corporate technology investment, and economy of scale effects. In reality, a great deal of technical change is led by the private sector, and is induced in response to market conditions, investment and expectations. In modeling terms, therefore, technical change really should be endogenous, i.e. dependent upon other parameters reflected within the model. Recent studies have tried to address some specific elements of induced technical change in limited ways. These studies are sufficient to suggest that the implications may be profound. For example, an improved representation of induced technical change may affect the optimal degree, nature, timing and distribution of abatement efforts. It is also clear that it may also affect international characteristics with respect to estimates of leakage, and international policy for example with respect to technology transfer. The main results available therefore suggest that the modeling of technical change is extremely important in deriving policy conclusions: the means by which technical change occurs may have strong implications for optimal environmental policy, and the economicmodelling thereof. Indeed, incorrect modelling of technical change can lead not only to wrong conclusions about the capacity to solve big environmental problems,but to policy recommendations that are in fact counterproductive. This pilot study seeks to clarify the economic issues relating toinduced technical change, reviews the modelling of technical change and what has been learnt from work during the 1990s, and discusses the nature of the possible policy implications. Following this introduction, the structure we follow is: 2. The importance of technical change. How important is it in the wider context? 3. Sources of technical change: classification and evidence. What is the evidence for induced vs. autonomous technical change? 4. Technology modeling in the classical models. How do the mainstream models account for technical change?
6
5. Explicit studies of induced technical change. What direct studies have been carried out and what do they tell us? 6. Incorporation of induced technical change in major macroeconomic & related models. What approaches are being explored and what is state of the art? Finally, we overview policy-related conclusions from our study.
2. The importance of technical change Technical progress holds the key to pollution abatement in the course of economic growth— and also to reducing the costs of pollution abatement. If we look at the history of pollutants like PM, lead in fuels, VOCs and more recently SO2 and NOx, technologies and fuels have been developed that are able to reduce pollution per unit of energy use by factors of ten to several thousand or more, such that even with rising energy consumption, we have seen dramatic reductions in emission levels in countries that have introduced policies to encourage pollution abatement. Moreover, the costs have often been far less than originally anticipated, sometimes becoming less than those of the fuels they displaced— the substitution of gas for coal in electricity generation and as a domestic and industrial fuel is a well-known example— or have been offset by efficiency gains elsewhere in the activity in question. In short, they are enabling us to reconcile energy use with a better environment with relatively little effect on energy costs. Will finding alternatives to fossil fuels to address climate change be an exception? Probably not, though considerable effort is still required to develop the alternative technologies— especially given the continuing difficulties with nuclear power. We know that a wide range of options is emerging, thanks to recent R&D, demonstration and market stimulationprogrammes: in a range of renewable energy technologies, for example, such as photovoltaics, wind (onshore and offshore), solar thermal, ocean based systems, geothermal, biomass gasification and combustion for power generation, hydrogen derived from renewable resources and fuel cells. There are also much discussed possibilities for the extraction of hydrogen from gas and coal, with the carbon (in the form of CO2) being used for enhanced oil and coal-bed methane recovery on closed, non-netCO2-emitting cycles. The IPCC (including the recent report of the team working on the Special Report on Emissions Scenarios) and many others have consistently suggested that a low carbon future can be reconciled with high and rising levels of world energy use in the long term through a shift to such technologies. Table 1 provides evidence on these points by showing the emissions intensities of technologies and practices for reducing emissions and environmental damage relative to the (polluting) alternatives they displace. Also shown are some indicators of the effects on costs, which are generally small, and sometimes, in retrospect, have turned out to be negative. . For each source and individual pollutant, the % proportion of emissions from available low polluting practices/technologies compared to the current average levels is shown in the second column. The third column shows the % proportion of marginal costs of low polluting options compared to the marginal cost of supply. The cost uncertainties related to renewable energy technologies and addressing climate change will be discussed shortly.
7
Table 1. Pollution intensities and costs for selected energy-related activities and pollutants Source and pollutant
% emissions ratio of lowpolluting to polluting practices (1)
Electricity generation from coal PM emissions
