REXS: یک مدل پیش بینی شده برای ارزیابی تاثیر مصرف منابع طبیعی و تغییرات تکنولوژیکی در رشد اقتصادی
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|10992||2006||50 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Structural Change and Economic Dynamics, Volume 17, Issue 3, September 2006, Pages 329–378
This paper describes the development of a forecasting model in the tradition of system dynamics. It is called Resource EXergy Services (REXS). The model simulates economic growth of the US through the 20th century and extrapolates the simulation for several decades into the next century. The REXS model differs from previous energy–economy models such as DICE and NICE [Nordhaus, W.D., 1991. The cost of slowing climate change: a survey. The Energy Journal 12 (1), 37–66] by eliminating the assumption of exogenously driven exponential growth along a so-called ‘optimal trajectory’. Instead, we suggest a simple model representing the dynamics of technological change in terms of decreasing energy (exergy) intensity and endogenously increasing efficiency of conversion of raw material and fuel inputs (exergy) to primary exergy services (‘useful work’). In our model, the traditional assumption of exogenous technological progress (total factor productivity) increasing at a constant rate is replaced by two learning processes based, respectively, on (i) cumulative economic output and (ii) cumulative energy (exergy) service (useful work) production experience. The initial results of simulation for the period 2000–2050 have significant implications for future trends in economic output. These implications are important for purposes of scenario analysis. The REXS modules are the focus of ongoing research. We discuss briefly the many possibilities for elaboration of each module to enrich the feedback dynamics, policy levers and post-scenario analyses.
Integrated assessment (IA) models are at the heart of efforts to assess policies and prospects for the future. The relationships between technological progress, economic activity and global environment are the focus of most of this research. Each model is designed to address different policy questions, for example, to quantify the potential costs of climate stabilisation policies such as the Kyoto protocol (Manne and Wene, 1994 and Weyant, 1999), or to assess our ability to meet future energy demands (Nakicenovic, 1993) and maintain future rates of economic growth (Gerlagh and van der Zwaan, 2003). Examples include ‘top–down’ (or macroeconomic) models, GEMINI-E3, MERGE, CETA, DICE and RICE (Bernard and Vielle, 2003, Manne and Wene, 1994, Manne and Richels, 2004, Peck and Teisberg, 1995, Nordhaus, 1993, Nordhaus, 1994a and Nordhaus, 1994b); and ‘bottom–up’ (energy system) models, MESSAGE, DEMETER or FREE (Messner, 1995, Fiddaman, 1998 and Gerlagh and van der Zwaan, 2003). Whether bottom–up or top–down, most share a common set of assumptions based on a neoclassical theory of economic growth applied both system dynamics (SD) and computable general equilibrium (CGE) models. We argue in this paper that the neoclassical theory fails to address critical issues relevant to integrated economic modelling and energy forecasting, namely the representation of technological change, and the role of materials and energy consumption in the economy. Technology has been included in IA models by a variety of methods: (i) carbon emissions reducing, (ii) cost reducing and (iii) output augmenting. In top–down models, technological progress is often only included as an output augmenting coefficient in the production function. DICE is a direct application of the standard neo-classical theory, and provides a typical example1. The top–down DICE model (Nordhaus, 1991)2 was one of the first integrated-assessment models of the economics of climate change, wherein the costs of mitigating climate change today were measured against the future “benefits” to be derived from economic growth. Gross output is given by a two-factor (capital and labour only) Cobb–Douglas production function. Results from his analysis led Nordhaus to claim that global warming might not actually be such a big problem. Some of the assumptions specific to the DICE model were challenged immediately3 (Cline, 1992, Tol, 1994 and Frankhauser, 1995). In particular, the model was criticised for failing to accurately represent either energy or technology interactions, and for its neoclassical formulation (Ayres, 2001). The level of technology is represented as an output augmenting multiplier (total factor productivity) which declines at an exogenous rate4 to an asymptotic value of zero. This ‘normative’ assumption about the future rate of technological progress implies in conjunction with slowing population growth and diminishing returns to capital investments, that economic growth stops within a century. For climate policy slow economic growth has considerable practical implications, as concomitant reductions in energy demand minimise the pressures on the climate (Fiddaman, 2002). Energy consumption is considered only as a consequence of growth, and not a driver of growth. Absent from the aggregate production function, the amount of energy consumed and the efficiency with which it is used have no impact on the level of gross output. In this context, it is assumed that energy consumption can decline to zero while economic activity continues. This is clearly misleading. It is now widely accepted that energy and materials are essential for industrial activity. The most widespread energy-augmented production function used in integrated models today (particularly of the general equilibrium variety) is the nested constant elasticity of substitution (CES) production function. The introduces a composite “capital–energy” good as a factor of production5 (Fiddaman, 1996 and van der Zwaan et al., 2002). However, despite the inclusion of energy and materials and regardless of the level of detail with which emissions reducing or efficiency enhancing technologies are modelled, all models relying on the CES (or the Cobb–Douglas function) to define gross output require the modeller to make normative assumptions of output augmenting technological progress (or total factor productivity) to drive the economy and maintain future levels of economic output. A closer look at the CES used in DEMETER (Gerlagh and van der Zwaan, 2003), but typical of those in other models makes this clearer,in which Q(t) is gross output, A(t) is the level of technological progress for capital and labour, Kc(t) is the capital stock, L(t) is the labour input, B(t) the technological progress of energy stocks,6F(t) is the fossil energy input and N(t) is the non-fossil energy input. In this version of the CES, there are two exogenous output augmenting multipliers A(t) and B(t). Moreover, the parameters α, χ and γ are constants describing the share of capital in the capital/labour composite, the elasticity of substitution between fossil and non-fossil energy use, and finally the elasticity of substitution between capital/labour on the one hand and the fossil/non-fossil composite, on the other hand (van der Zwaan et al., 2002). With technological and structural change the assumption of constant elasticities of output and substitution is highly questionable (Sylos Labini, 1995). We shall come back to this point, when discussing the choice of aggregate production function in REXS (Section 2.5). Efforts to endogenise specific technology dynamics have been most prevalent in bottom–up models. Endogenous rates of efficiency improvements and cost reductions are represented for a rich set of specific energy technologies and learning-by doing dynamics (Nakicenovic and Riahi, 2002). In the MESSAGE model, Messner and Strubegger (1995) introduced learning curves to determine the rate of decrease of costs, emissions and market penetration constraints as investments and installed capacities accumulate (Gerlagh and van der Zwaan, 2003). Modifications of the top–down DICE model to incorporate energy flows, resulting in the NICE and FREE system dynamics models (Fiddaman, 1996, Fiddaman, 1997 and Fiddaman, 1998) include key changes to the capital growth loop and energy systems representation. Indeed, many top–down (CGE) models, such as have since been modified to incorporate a more detailed energy systems representation, typical of bottom–up models. Both NICE and FREE include stocks, flows, non-linearities and disequilibrium feedback mechanisms.7 They also and require a certain degree of rationally bounded decision making to control certain parameters. These modifications have considerably enriched the feedback dynamics. Cost reductions due to cumulative production experience (learning and scale economies) provide scope for lock-in of fossil fuels vis-à-vis alternatives. These models have effectively endogenised selected components of technological progress. The energy-intensity of the capital–labour composite is controlled by the relative marginal returns to energy, capital and labour in the short run (on the basis of learning-by doing dynamics). In the long run, it is controlled by an exogenous assumed rate of productivity increases and cost reductions in the energy sector. Simulation results generally imply stronger emission reductions in the near term. As production experience accumulates, technological progress for non-carbon energy sources is more rapid (Fiddaman, 1996 and Grubler and Messner, 1998). The abatement costs (or shadow carbon prices) simulated using NICE or FREE are typically higher than those predicted by DICE and the range of uncertainty is considerably wider (∼US$ 15–135/tC optimal carbon tax in 2105) (Fiddaman, 1996). Nevertheless, regardless of the level of detail with which emissions reducing or efficiency enhancing technologies are modelled, whether the models are bottom–up or top–down, most maintain and require assumptions of output augmenting technological progress (or total factor productivity) to drive the economy and maintain future levels of economic output. This is the focus of our critique and the main topic of this paper. The essential difficulty can be restated: if one cannot explain the technological progress that took place over the past century, except as a residual, how can one expect to account for future technological progress except by assuming it? We argue that assumptions about technological progress are required because of misspecification of the interactions between technology and the efficient use of energy, and the effects this progress has in driving output growth. Existing energy–economy models place emphasis on showing the consequences of technological change within energy systems, not on the specific effects that flows of natural resources may have on productivity. If the dominant contribution to observed economic growth is attributable to a ‘fudge factor’, then it is likely that important mechanisms are missing from the current models. Whether formulated as total factor productivity or ‘technological progress’ the assumption of an exogenous output augmenting multiplier is troublesome. The solution we propose differs radically from other models discussed in the literature. It is called Resource EXergy Services (REXS). The model has been calibrated using historical data from the US that spans a century, and is capable of simulating the economic growth of the US accurately throughout the 20th century. The REXS model eliminates any assumption of exogenously driven exponential growth along a so-called ‘optimal’ trajectory. In the REXS model, the traditional assumption of exogenous technological progress (total factor productivity) increasing at a constant rate is replaced by two learning processes. In the first, production experience drives down the energy and materials intensity of output. In the second, experience gained in supplying energy to the economy acts to increase the efficiency with which energy is supplied to the economy in useful form, which in turn drives economic output hence energy demand and so on. The main innovations within the REXS model, discussed in the following sections of this paper, include, (i) the use of an alternative measure “exergy services” or ‘useful work’ to describe the productive inputs derived from materials and energy into the economy, (ii) redefinition of technological progress as a measure of the efficiency of conversion of energy into useful form (exergy services), (iii) and consequently the selection of an alternative to the neoclassical (energy-augmented) production functions, capable of adequately forecasting past and future rates of economic output in this modified energy–technology–economy system. The model results we present are not directly comparable with other global integrated assessment models, as we consider only the macroeconomic components of such models.8 The purpose of the REXS model is to highlight the importance of the relationship between the way that energy is supplied to the economy, the effects of technological progress to improve natural resource use efficiency and the effect of such efficiency improvements on productivity. However, we suggest that most models could be modified as we suggest. We begin by discussing the inadequacies of the neoclassical theory and by providing evidence of alternative growth engines that are potentially capable of explaining much of the observed past growth and by explaining the meaning of exergy and ‘useful work’ or ‘exergy services’ as we have chosen to name them. This sets the basis for a discussion of our ‘interpretation’ of the process of technological progress and our method of measuring it, and a complete description of the REXS model. We will complete the paper with a discussion of the policy implications of the preliminary results and suggestions for future versions of the REXS model.
نتیجه گیری انگلیسی
The REXSF model is a relatively simple quasi-endogenous system dynamics model capable of providing plausible scenarios for future economic growth. Output augmenting technologies are endogenously driven, while the rate of natural resource output intensity (R/GDP), capital investment-depreciation and labour dynamics are exogenously determined, with reference to scenario assumptions concerning future policy decisions. It contains comparatively few parameters and can be calibrated using empirical data, unlike models that involve measures of human capital to parameterize technological progress. The REXS model does this in another way: by explicit recognition that natural resource consumption is both a cause and effect of past economic growth, and that the rate of the improvement of the aggregate thermodynamic efficiency of exergy conversion of an economy is a realizable measure of its technological progress. Cumulative production and output provide operational measures to control the long-run dynamics of energy intensity decline (dematerialisation) and end-use efficiency improvements. The alternative (LINEX) that we propose to the standard production functions (Cobb–Douglas or CES) used in the majority of integrated economy-environment models provides important insights into the physical underpinnings of economic growth and has particular qualities that make it suitable for long-term forecasting. Of particular relevance under structural and technological change, is the lack of any assumption that (a) the factor productivities are constant, and (b) equal to their factor share in the national accounts. Indeed, if we look at the factor productivities (output elasticities) we see that since the early part of the century (1920) labour has the lowest value (∼0.1–0.2), exergy services are intermediate and capital takes the lions’ share. These results are similar to those originally presented by Kuemmel et al. (1985). However, despite the relatively low and declining productivity of labour, empirical evidence shows that wage rates in the US have been increasing over the last century in proportion to GDP, and the real costs of energy and materials has been declining (Nordhaus, 1994a and Nordhaus, 1994b). How can we understand this apparent contradiction with our findings? The answer lies in understanding that the LINEX production function reveals the hidden dynamics of productivity, and an economy that is quite far from equilibrium. One of the fundamental assumptions of neoclassical economics has not been satisfied in the US economy (typical of others, i.e. Japan, Germany, UK) since the early 1920s. Under pressures of cost minimisation industrial economies have been driven to substitute for expensive human labour, cheaper (and more powerful) exergy services. This substitution (together with economic growth) increases the Y/L and K/L ratios thus suggesting increasing labour productivity through capital deepening. This implication justifies ever higher wage rates (typically growing in proportion to output), leading to further substitution of capital and exergy services for labour and increasing unemployment. Indeed, as we have shown, labour is not as productive as the capital–exergy service combination. Much of the output growth experienced over the last 100yrs is the result of improvements in the way that natural resources are used to provide ‘useful work’. Yet, there is no direct measure of the “price of work” (except in the case of electricity) and therefore it is not possible to estimate a return to work (analogous to returns to labour and capital). Indeed, no one has been trying. Lacking this important measure, productivity improvements are attributed to aggregate labour and capital. However it is clear that cheap exergy/capital combinations will continue to replace costly labour/capital combinations far into the future. The policy implications are many. Firstly exergy services derived from natural resources are an essential factor of production, as opposed to the exergy flux per se, which includes waste flows that are more likely to hinder growth. Secondly, if objectives to dematerialise are not to slow output growth, the technical efficiency with which natural resource flows are used must increase. The rate of increase should match past improvements. Efforts to improve resource use efficiency must focus on the most ‘inefficient’ and wasteful activities, many of which have grown unchecked. For example, electric heat, personal transport by car, and ‘labour saving’ electrical (and gasoline powered appliances), have proliferated in recent years, even though the consumer price of energy (exergy) services is no longer declining. The role of labour in the US economy has changed over the past century, becoming increasing supervisory in nature (Beaudreau, 1998). Through investments in automated capital and consumption of exergy services output growth has boomed and the output intensity of labour (Y/L) has fallen dramatically. Lacking the insight provided by consideration of the role of exergy services, these developments have been interpreted simply as increases in the productivity of labour. As a consequence, governments have imposed taxes on employed labour in proportion to the perceived productivity increases. The same governments have also been responsible for providing significant (indirect) subsidies to invest in capital and to reduce the cost of the provision of exergy services. This is agreeable for businesses, who understand the real cause of productivity increases. But governments should understand that the consumption of natural resources to provide exergy services has significant negative externalities, and even poses a threat to future generations (e.g. pollution, ill-health). Hence efforts to increase employment are likely to have considerable positive externalities. Governments should tackle this issue in two ways. Firstly by transferring the tax and regulatory burden from labour to natural resource consumption. This would help to correct market failures and provide a greater incentive to industries to invest in more exergy efficient technologies, while also encouraging them to consider how investment in labour may assist them in becoming more efficient and productive. Secondly, accepting that the decline in demand for physical services provided by labour is likely to increase as technological advances increase even further the productivity of the automated capital–exergy service combination, governments and individuals must make significant efforts to provide a labour force of higher quality, capable of supervising complex automated processes. The past century has witnessed important structural changes to all aspects of the socio-economic system and in particular in the quantity with which energy and materials are supplied and the ways in which it is processed and used productively. The abrupt changes in some of the empirical time series discussed in this article are testament to this. However, in REXS the growth of technical efficiency is modelled as a continuous process. This is consistent with the observation that the aggregate technical efficiency f, of the US has been a simple increasing function of time. Future modifications of the model to improve the feedback dynamics may include: (a) Disaggregation of primary fuel mix and lock-in (inertia) price effects (supply side); (b) Distinction between carbon and non-carbon fuels and carbon tax effects; (c) Allowing for the influence of declining fossil fuel reserves on exergy price; (d) Disaggregation of technical efficiency by exergy carrier (demand side); (e) Possible endogenous controls of the R/GDP ratio. These changes would permit greater flexibility to control supply and demand side dynamics and allow us to explore various energy–economy–efficiency scenarios. We are aware the model has been applied thus far only to a single country (the US). We cannot yet discuss climate effects, taxes and utility, or extend the specific conclusions derived from results that were possible for global models. However, we do suggest that the dynamics represented in the REXS model are likely to be generally applicable to energy–economy models of individual countries, regions and the World. The inclusion of exergy service dynamics removes the need for exogenous assumptions of continuous exponential growth, reveals the full importance of natural resources in the economy and provides an operational representation of technological progress, suitable for the generation and analysis of scenarios of the future.