1. FISIOLOGÍA REPRODUCTIVA DEL DELFÍN MULAR (Tursiops truncatus)
1.2. Fisiología reproductiva del macho de delfín mular
In September 2015, Brazil submitted the Intended Nationally Determined Contribution (INDC) towards achieving the objective of the United Nations Framework Convention on Climate Change (UNFCCC), for the COP21 meeting in Paris (Brazilian Government, 2015). According to the INDC, Brazil committed to:
• Reduce GHG emissions by 37% below 2005 levels in 2025 (target)
• Reduce GHG emissions by 43% below 2005 levels in 2030 (indicative)
The Brazilian commitment builds on the current low carbon configuration of its energy sector (as explained in the previous section), and an already successful plan for reducing emissions produced by land use changes (IEA and OECD, 2013, p. 318). Figure 8.7 shows how GHG emissions in Brazil decreased by 33% between 2005 and 2014 (29% if is measured between 2000 and 2014). The decrease in emissions, however, is not balanced accross sectors:
8.4 Power sector emission in Brazil and the INDC 155
most of the reduction comes from land use changes. In all the other sectors, emissions increase persistently over the same period 2000-2014, with waste and energy having the larger percentage increase, 76% and 61% respectively. If the INDC target is going to be achieved, then other complementary mechanisms will have to be implemented, especially to address the persistent rise in emissions from the agriculture and energy sectors.
Figure 8.7 Brazilian GHG emissions between years 2000 and 2014, by sector. Data from Sistema de Estimativa de Emissoes de Gases de Efeito Estufa˜ (SEEG, 2016).
An ensemble with 120 FTT:Power simulations is presented below, to analyse the potential emission pathways of the Brazilian power sector under different scenarios of hydropower availability, in the context of the INDC commitments. Using the resource distributions described in section 6.6 (see figure 6.6), the cost supply curve range of hydroelectricity in Brazil is sampled,4 so that each scenarios has a different level of hydropower availability. All the policy instruments are applied on the scenarios (which corresponds to the policy set “All”), using he same assumptions described in section 8.3. The emission trajectories of the 120 scenarios are aggregated in figure 8.8. For reference purposes, the BAU trajectory is plotted in black at the top of the chart.
At the extreme left of figure 8.8, are the emissions of the power sector in 2005, which is the base year of Brazil’s INDC. If emissions were to be cut proportionally among sectors, then
4The sampling method used is latin hypercube sampling (LHS), which guaranties a proportional sampling
from the entire distribution range, and therefore is preferred over the traditional Monte Carlo Sampling method (Helton and Davis, 2003)
an emission reduction of 37% would be required by 2025 in the power sector, and that is indicated by the brown dashed horizontal line. In the context of the FTT:Power scenarios analysed in this work, based on the policy sets described in section 5.2, it is not possible to decrease 37% of emission from the power sector, even in the scenarios of high hydro resources availability, let alone the scenarios of low hydropower availability.
Figure 8.8 The cost supply curve for hydroelectricity in Brazil is sampled, to produce 120 scenarios for the power sector. The shading represents the frequency in the emission trajectories of FTT:Power outputs, based on the distribution of the cost supply curve for hydroelectricity from the NER module. All the scenarios are based on the policy set “All” (except the black line), which assumes that carbon pricing, subsidies, feed-in-tariffs and regulation are in place (for the details about each policy instrument, please refer to section 5.2). The solid red lines represent emission from scenarios produced with the upper and lower limits of the cost supply curve. The solid blue line corresponds to the emissions trajectory from the scenario produced with the most likely cost supply curve. The horizontal dashed brown line indicates a 37% reduction in emissions, in comparison with 2005 level, and the vertical dashed brown line indicates the year 2025, which is the target year for the INDC of Brazil. The black line corresponds to the emissions trajectory from the BAU scenario, inserted for reference purposes only, because is not part of the “All” policy set. In the BAU scenario, no decarbonisation policies are implemented.
8.4 Power sector emission in Brazil and the INDC 157
The emission trajectories corresponding to the range between the most likely and the upper limit of the cost supply curve of hydroelectricity in Brazil (the upper range of hydropower resource availability), are concentrated between the bottom red and blue lines of figure 8.8, which are overlapping almost entirely. This means that for high availability of hydroelectric resources, there is almost no difference in terms of power sector abatement. This is not surprising: if the total demand for hydroelectricity is matched by the supply, more availability of hydro resources do not provide further decarbonisation potential.
In contrast, the emission trajectories associated with the range between the most likely and the lower limit of the cost supply curve (the lower range of hydropower resource availability), are distributed between the blue and upper red lines of figure 8.8. The large range of emissions associated with the lower range of the hydro cost supply curve, indicates the importance of hydroelectricity in Brazil to maintain its low carbon energy matrix. It is important to notice that, given the exponential distribution of resources, the probability associated with higher emission trajectories are expected to be low. However, despite the low probabilities associated with the high emission scenarios, it is important to understand the underlying risks, which may increase over time due to the effects of climate change in Brazil.
The low carbon intensity of the Brazilian power sector hinder further reductions of GHG emissions, which are required for achieving the international commitments stated by the INDC. As the emission trajectories presented in figure 8.8 show, Brazil has limited decar- bonisation possibilities within the power sector, given the large amount of hydroelectricity that is already in place. In order to further reduce power sector emissions, given the current technological lock-in, stringent decarbonisation measures will be required, e.g., regulation on the end-use sector, or early scrapping of carbon intensive power stations (IEA and OECD, 2012a).5
Given the existing limitations related to further decarbonise the Brazilian power sector, it is necessary to go beyond traditional sectoral approaches, and look at the system as a whole. In order to look for effective decarbonisation policies, aligned with the INDC commitment, it is absolutely necessary to understand the complex linkages between the food, water and energy sectors in Brazil, usually referred as the Nexus (Bazilian et al., 2011). Future changes in the global climate could potentially have detrimental impacts on the land cover and biodiversity in Brazil, with implications on agriculture, cattle and food production, water for electricity production, and consequently on emissions. As can be seen in the data from SEEG, presented in figure 8.7, the land use and agriculture sector can play an important role in further decarbonise the Brazilian economy.
8.5
Conclusions
This chapter analyses the impact of hydropower resources availability on the performance of decarbonisation policies in Brazil. Using the database of energy resources of the NER module, introduced in chapter 6, 15 policy sets are analysed under extreme scenarios of hydropower availability. Carbon pricing, subsidies, feed-in-tariffs, direct regulation, and any possible combination of them, are compared in terms of their policy efficiency. The policy set “All”, that includes a combination of all the policy instruments, is analysed in detail. 120 emission scenarios are aggregated and contrasted with the Intended Nationally Determined Contribution submitted by Brazil for the COP21. From this analysis, the main conclusions obtained are:
• The large hydropower-dependence of the Brazilian power sector hinders the efficiency of strong decarbonisation policies, particularly under scenarios of low availability of hydropower resources. As figure 8.4 shows, large differences in policy efficiency are exhibited between the low and high hydro availability scenarios, for all the policy sets. Similar to the results presented in chapter 7, regulation is the single policy instrument with the best decarbonisation performance.
• If Brazil were to apply its emission reduction target of 37% (below 2005 levels in 2025) uniformly across sectors, then the policy sets used in this work are not able to deliver the required emission reductions, even in the scenario with the highest availability of hydropower resources. In the scenario with the lowest availability, the gap between the target and the expected emission trajectories increases (see figure 8.8).
• Over the past 15 years, energy sector emissions have been increasing steadily in Brazil (see figure 8.7). In order to achieve its INDC target, Brazil cannot rely on the decarbonisation of the power sector, which has a limited decarbonisation potential. Therefore, comprehensive decarbonisation policies are required, involving several sectors, especially energy, agriculture, and land use.
Chapter 9
Learning
9.1
Chapter Summary
According to the International Energy Agency, investments in the order of USD 140 trillion are required to achieving a low-carbon energy sector by 2050 (IEA and OECD, 2012a, p. 135). The decarbonisation of the energy sector requires a profound and rapid technological transformation, and governments are required to provide well suited economic and legal frameworks for private and public investment to be allocated in low carbon energy alternatives (IPCC, 2014b). In order to produce effective policies to stimulate low carbon investment, feasible economic and technological scenarios for the future have to be analysed. Such analysis requires a good understanding of the technological landscape, and its potential evolution driven by public and private investment. It is crucial, therefore, to understand how technological change works, and how it can be modelled.
In line with the aforementioned challenges, this chapter provides an introduction to the subject of technological change and learning, how is modelled in FTT:Power (in contrast with some classical approaches), and presents future decarbonisation scenarios under uncertain technological change (in the form of extreme learning rates). The chapter structure is divided as follows:
• First, section 9.2 provides an introduction to the subject of technological change and learning modelling.
• Second, section 9.3 addresses some of the core areas of endogenous technological change modelling, including learning by doing, spillovers and path dependency.
• Third, section 9.4 introduces some of the main sources of uncertainty in the process of learning, and how they are addressed (or not) by the modelling community.
• Fourth, section 9.5 links technology clusters and learning, and show how the link is modelled in FTT:Power.
• Fifth, section 9.6 discusses the difficulties of measuring learning rates, and conse- quently, the complexities associated with modelling the process of technological change.
• Sixth, section 9.7 explains how learning curves are implemented in FTT:Power, and provides uncertainty intervals for each of the learning coefficients used in the model. These learning intervals were obtained from a thorough literature review, which is presented in the Appendix section C.1.
• Seventh, section 9.8 presents an analysis of the impact that uncertainty in the values of the learning rates has on the emission trajectories of the power sector. Decarbonisation scenarios of the power sector are presented, using extreme learning rates, and different configurations of the system regarding the capability of the grid to incorporate renew- able energies. Based on these decarbonisation scenarios, it is concluded that the impact of the grid flexibility on emission reductions is larger than the impact of uncertainty in the learning rates.
• Eighth, section 9.9 concludes with the highlights of the modelling exercise, and some policy recommendations.
9.2
Introduction
Historically, economic models have included technological change and innovation as an important driver of economic growth (Aghion and Howitt, 1992; Romer, 1990; Schumpeter, 1934; Solow, 1956). Early representations of technological change were mostly exogenous and, therefore, unresponsive to drivers such as public policy, R&D investment or regulation. The cornerstone of exogenous technological change in macroeconomic was provided by Solow, who argued that the unexplained difference in productivity coming from its econo- metrics analysis of US data was due to technological progress (Solow, 1956; Verspagen, 1992). From there, several models focused on human capital formation as the key factor for explaining technological change and economic growth (Phelps, 1966; Shell, 1967; Uzawa, 1965).
9.2 Introduction 161
The endogenisation of technological change as a factor of economic growth started with the inclusion of ‘externalities’ to explain increasing returns to scale, such as knowledge spillovers (Romer, 1990), international interdependence through trade (Grossman and Helpman, 1994) or entrepreneurial innovation (Audretsch, 1995; Audretsch, David B. and Keilbach, Max C., 2004). Since then, numerous attempts to model technological change as an endogenous factor of economic growth have been produced, many of which can be found on the vast amount of existing literature reviews (Kahouli-Brahmi, 2008; Kohler et al., 2006; Rubin et al., 2015; van der Zwaan and Seebregts, 2004). In many of the existing models, human capital is considered the main input in the innovation process (Verspagen, 1992). Human capital is produced by both private firms in the form of blueprints (Aghion and Howitt, 1992) and from R&D in the form of general knowledge (Grossman and Helpman, 1994; Romer, 1990). In both cases, the link with research and innovations is the key factor for technological progress.
Building on Solow’s exogenous growth model, Kenneth Arrow provided an alternative endogenous theory for explaining the gap between increase in per capita income and the increase in the capital-labour ratio (Arrow, 1962). In its remarkable work about ‘learning by doing’ (or LBD), Arrow provided a simple representation of the complex relation between experience and productivity. The empirical counterpart to Arrows’ theoretical contribution is the ‘learning curve’ or ‘experience curve’, a concept formally introduced by the Boston Consulting Group (BCG) in 1966, but discovered by Wright a few decades earlier.1 BCG was performing a cost analysis for a major semiconductor manufacturer (Reeves et al., 2013), when they found that the unit production costs felt by a predictable amount (typically 20 to 30 percent in real terms), for each doubling of ‘experience’, or accumulated production volume. That fractional decrease in cost as a function of the production is denominated ‘learning rate’. Three decades earlier, T. P. Wright found a similar pattern followed by the airplane manufacturing industry, something that he described as ‘variation of cost due to production experience’ (Wright, 1936).
The seemingly simple observation made by Wright and the BCG, that cost per unit output declines as production grows, has been observed in several manufacturing industries ever since, including electronics, machine tools, system components for electronic data process- ing, papermaking, aircraft, steel, apparel, and automobiles (IEA, 2000). The concepts of
1In general, the terminology ‘learning curves’, ‘experience curves’ and ‘progress curves’ are used indis-
tinctively. However, some subtle differences can be found in the literature. For instance, Dutton and Thomas differentiate ‘experience curves’ from ‘progress curves’ in their scope. The former represents average produc- tion cost of multiple manufacturers, whereas the latter represents production costs at the firm level (Dutton and Thomas, 1984; Weiss et al., 2010). In this work, the three terms are considered equivalent.
learning and experience curves are used indistinctively to represent the empirical log-linear relationship between production cost and cumulative experience (using typically cumulative installed capacity or cumulative sales as a proxy of experience. See section 9.3 below for more details).
Whilst there is vast evidence of the positive influence of knowledge (typically measured using proxies such as R&D expenditure, education or patents) in productivity and economic growth, the magnitude of such influence is still unclear (Griliches, 1994; Hu et al., 2005; Mason et al., 2012; Samaniego, 2007). Despite the lack of consensus in the magnitude, knowledge is considered by many economists as capital stock with a positive influence in productivity and increasing returns to scale. This is valid in both cases, when knowledge is represented by R&D or blueprints from public and private companies (classical growth models), and when knowledge is characterised as experience (LBD perspective). However, the process of acquisition of knowledge (learning) is intrinsically complex and difficult to be measured. For that reason, modelling technological change endogenously presents a series of barriers, especially in the context of optimisation and equilibrium. For instance, learning by doing makes the problem of minimising total energy system costs non-convex, giving rise to multiple equilibria and attendant instability of models’ numerical solution (Sue Wing, 2006). Moreover, knowledge has increasing marginal productivity (Romer, 1986), which produces increasing returns to scale and path dependency, leading to technological lock-in (David, 1985). All of these factors hinder the incorporation of endogenous technological change in economic (top-down) and energy (bottom-up) models.