11.5 Technology and the costs of mitigation
11.5.1 Endogenous and exogenous technological development and diffusion
A major development since the TAR has been the treatment of technological change in many models as endogenous – and therefore potentially induced by climate policy – compared to previous assumptions of exogenous technological change that is unaffected by climate policies (see glossary for definitions). This section discusses the effect of the new endogenous approach on emission permit prices, carbon tax rates, GDP and/or economic welfare, and policy modelling (Chapter 2, Section 2.7.1 discusses the concepts and definitions, and Chapter 13 provides a broader discussion of mitigation and technology policy choices).
The TAR reported that most models make exogenous assumptions about technological change (9.4.2.3) and that there continues to be active debate about whether the rate of aggregate technological change will respond to climate policies (7.3.4.1). The TAR also reported that endogenizing technological change could shift the optimal timing of mitigation forward or backward (8.4.5). The direction depends on whether technological change is driven by R&D investments (suggesting less mitigation now and more mitigation later, when costs decline) or by accumulation of experience induced by the policies (suggesting an acceleration in mitigation to gain that experience, and lower costs, earlier). Overall, the TAR noted that differences in exogenous technology assumptions were a central determinate of differences in estimated mitigation costs and other impacts.
Table 11.13 lists the implications for modelling of exogenous and endogenous technological change and demonstrates the challenges for research. The table shows that, at least in their simplified forms, the two types of innovation processes potentially have very different policy implications in a number of different dimensions.
Table 11.13: Implications of modelling exogenous and endogenous technological change
| | Exogenous technological change | Endogenous technological change |
|---|
| Process | Technological change depends on autonomous trends | Technological change develops based on behavioural responses, particularly (a) choices about R&D investments that lower future costs; and (b) levels of current technology use that lower future technology cost via learning-by-doing |
| Modelling implications |
|---|
| Modelling term | Exogenous | Endogenous / induced |
| Typical main parameters | Autonomous Energy Efficiency Index (AEEI) | Spillovers to learning / return to R&D / cost of R&D / Learning rate |
| Optimization implications (note: not all modelling exercises are dynamically optimized) | Single optimum with standard techniques | Potential for multiple-equilibria; unclear whether identified solutions are local or global optima |
| Economic/policy implications |
|---|
| Implications for long-run economics of climate change | Atmospheric stabilization below approximately 550 ppm CO2 likely to be very costly without explicit assumption of change in autonomous technology trends. | Stringent atmospheric stabilization may or may not be very costly, depending on implicit assumptions about responsiveness of endogenous technological trends. |
| Policy instruments that can be modelled | Taxes and tradable permits | Taxes and tradable permits as well as R&D and investment incentives / subsidies |
| Timing implications for mitigation and mitigation costs associated with cost minimization | Arbitrage conditions suggest that the social unit cost of carbon should rise over time roughly at the rate of interest. | Learning-by-doing implies that larger (and more costly) efforts are justified earlier as a way to lower future costs. |
| ‘First mover’ economics | Costs with few benefits | Potential benefits of technological leadership, depending on assumed appropriability of knowledge |
| International spillover / leakage implications | Spillovers generally negative (abatement in one region leads to industrial migration that increases emissions elsewhere) | In addition to negative spillovers from emission leakage / industrial migration, there are also positive spillovers (international diffusion of cleaner technologies induced by abatement help to reduce emissions in other regions) |
The role of technology assumptions in models continues to be viewed as a critical determinant of GDP and welfare costs, and emission permit prices or carbon tax rates (Barker et al., 2002; Fischer and Morgenstern, 2006). These analyses cover large numbers of modelling studies undertaken before 2000 and regard the treatment of technology as influential in reducing costs and carbon prices, but find that the cross-model results on the issue are conflicting, uncertain and weak. Since the TAR, there has been considerable focus on the role of technology, especially in top-down and hybrid modelling, in estimating the impact of mitigation policies. However, syntheses of this work tend to reveal wide differences in the theoretical approaches, and results that are strongly dependent on a wide range of assumptions adopted (Barker et al., 2006a; Stern, 2006), about which there is little agreement (DeCanio, 2003).
The approaches to modelling technological change (see Section 2.7.2.1), include (1) explicit investment in research and development (R&D) that increases the stock of knowledge, (2) the (typically) cost-free accumulation of applying that knowledge through ‘learning-by-doing’ (LBD); and (3) spillover effects. These approaches are in addition to simple analyses of sensitivity to cost assumptions, especially when technological change is treated as exogenous. There have been many reviews (see Clarke and Weyant, 2002; Grubb et al., 2002b; Löschel, 2002; Jaffe et al., 2003; Goulder, 2004; Weyant, 2004; Smulders, 2005; Grübler et al. 2002; Vollebergh and Kemfert, 2005; Clarke et al., 2006; Edenhofer et al., 2006b; Köhler et al., 2006; Newell et al., 2006; Popp, 2006b; Sue Wing, 2006; Sue Wing and Popp, 2006). One feature that emerges from the studies is the considerable variety in the treatment of technological change and its relationship to economic growth. Another is the substantial reductions in costs apparent in some studies when endogenous technological change is introduced, comparable to previously estimated cost savings from ad hoc increases in the exogenous rate of technological change (Kopp, 2001) or in the modelling of advanced technologies (Placet et al., 2004 p. 5.2 & 8.10).
This section reviews the effect of endogenizing techno-logical change on model estimates of the costs of mitigation. It follows the majority of the literature and takes a cost-effectiveness approach to assess the costs associated with particular emission or cumulative emission goals, such as post-2012 CO2 reduction below 1990 levels or medium-term pathways to stabilization.
The review shows that endogenizing technological change – via R&D responses and learning-by-doing – lowers costs, perhaps substantially, relative to estimates where the path of technological change is fixed from the baseline. The degree to which costs are reduced hinges critically on assumptions about the returns from climate change R&D, spillovers (across sectors and regions) associated with climate change R&D, crowding-out associated with climate change R&D, and (in models with learning-by-doing) assumed learning rates. Table 11.14 shows the policies that have been modelled to induce technological change, and how they have been introduced into the models.
Table 11.14: Technology policies and modelling approaches
| Policies | Modelling approach | Key points for measuring costs |
|---|
R&D in low-GHG products and processes from: • Corporate tax incentives for R&D (supply-push R&D) • More government-funded R&D (supply-push R&D) | • Explicit modelling of R&D stock(s) that are choice variables, like capital, and enter the production function for various (low-carbon) goods. • R&D policies can be modelled as explicit increases in R&D supply or subsidies for the R&D price. | The assumed rate of return from R&D, typically based on an assumption that there are substantial spillovers and that the rate of return to R&D is several times higher than conventional investment at the margin due to spillover. Another important point is the assumed cost of R&D input, which may be high if it is taken from other R&D (crowding-out) |
Learning-by-doing: • Purchase requirements or subsidies for new, low-GHG products • Corporate tax incentives for investment in low-GHG products and processes | More production from a given technology lowers costs. | Rate at which increases in output lowers costs and long-run potential for costs to fall. |
The policies are in two groups: effects through R&D expendi-ture, and those through learning-by-doing.Unfortunately, our empirical understanding of these phenomena over long periods of time is no better than our ability to forecast exogenous rates of change. As Popp (2006b) notes, none of the ETC models he reviews use empirical estimates of technological change to calibrate the models because, until recently, there were few empirical studies of innovation and environmental policy. So although we are confident that mitigation costs will be lower than those predicted by models assuming historically-based, exogenous rates of technological change, views continue to differ about how much lower they will be.