The rebound effect: An evolutionary perspective

EC O LO GIC A L E CO N O M ICS 6 7 ( 2 00 8 ) 5 2 6 –5 37 a v a i l a b l e a t w w w. s c i e n c e d i r e c t . c o m w w w. e l s e v i e r. c o m / l o c a t e / e c o l e c o n METHODS The rebound effect: An evolutionary perspective F. Ruzzenentia , R. Basosia,b,⁎ a Center for Complex System Investigations, University of Siena, Italy Department of Chemistry, University of Siena, Italy b AR TIC LE D ATA ABSTR ACT Article history: The rebound effect presents a major flaw in to energy conservation policies that aim to Received 11 October 2007 reduce energy consumption through energy efficiency development. Economics and energy Received in revised form 7 July 2008 related disciplines have thus far developed tools to measure such a phenomenon. This Accepted 1 August 2008 paper attempts to explain this seeming paradox using a thermodynamic-evolutionary Available online 1 September 2008 theoretical framework in addition to the traditional economic approach. We here propose that evolutionary systems, such as biological or even economic systems, may rearrange Keywords: themselves in a more complex fashion under the pressure of an increasing flux of energy, Rebound effect driven by the higher conversion rate of greater efficiency. Higher complexity, due to a Energy efficiency greater energy density rate, counteracts the positive effects of energy efficiency. We Energy density rate investigated this hypothesis in the context of the road freight transport system and the Complexity productive structure. The qualitative analysis in this paper, further substantiated by figures, Evolutionary approach provides a link between the dynamics of production patterns and the effect of efficiency in Globalization the light of the macro-economic effects of increased energy demand. The analysis departs Road freight transport system from a rigorous investigation of the actual energy efficiency evolution in the road freight transport system to develop through a survey of the subsequent worldwide economic revolution in the production system. It is then shown how outsourcing, the key feature of globalization, can be identified as the main source of traffic density growth. Finally, four paradigms are used to stress how the shift in the production system must be considered a leap in structural complexity that consequently serves to increase the frequency of components’ interactions. © 2008 Elsevier B.V. All rights reserved. 1. Introduction The rebound effect is portrayed by economic literature as a price-adjusting phenomenon that counterbalances, partially or thoroughly, the conservation effects expected by the adoption of a more efficient technology (Khazzoom, 1980; Brookes, 1990; Saunders, 1992). It involves an upswing of new costs brought about by the introduction of new energy conversion technology. Economic theory, however, seems to focus solely on the problem of estimating the size of the effect, i.e. the percentage of savings offset by the increased demand (Greening et al., 2000; Berkhout et al., 2000; Dimitropoulos, 2007). In the present analysis, we address the rebound effect from evolutionary and thermodynamic perspectives using an approach first envisioned by Alfred Lotka in the field of life sciences (Lotka, 1956). The aim of this analysis is therefore to provide unique insight on the issue in order to achieve a novel explanatory framework of the phenomenon. We seek this new framework due to our conviction that phenomena exhibiting significant analogies, ⁎ Corresponding author. Center for Complex System Investigations, University of Siena, Italy. E-mail address: [email protected] (R. Basosi). 0921-8009/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.ecolecon.2008.08.001 EC O L O G IC A L E C O N O M IC S 6 7 ( 2 0 08 ) 52 6 –5 37 albeit belonging to different fields, can be approached using a broader theory that is general in terms. There are indeed similarities between economical and biological systems. One important analogy, particularly relevant to the present research, is the relationship between higher efficiency and higher energy level: systems with a better energy conversion rate also display a higher energy density rate. If the positive relationship between energy efficiency and energy density is a constant of thermodynamic evolutionary systems then the analogy can be broadened to a third feature: complexity (Odum, 1996). It is considered fundamental knowledge of evolutionary biology that greater complexity is linked to greater efficiency. Evolution thus awards the highest efficiency achieved to the highest structural complexity throughout the system (Alberts et al., 1994). Furthermore, it is well known that greater complexity entails higher energy costs and thus, the greater the complexity, the higher the energy density rate. It is therefore possible that the formation of a more complex structure may offset the conservation effect brought about by a more efficient technology. The explanation of the rebound effect would rest in the complexity leap of the system following the advent of new technology. More specifically, the rebound effect addressed here is the macro rebound effect or wide-economy rebound effect, a specific type of rebound effect that is considered most crucial in determining energy consumptions (Sorrell and Dimitropoulos, 2007). The macro rebound effect underlies a long-term adjusting mechanism affecting factors markets. As stated by Sorrell and Dimitropoulos: “a fall in the real price of energy services [that] will reduce the price of intermediate and final goods throughout the economy, leading to a series of price and quantity adjustments, with energy-intensive goods and sectors gaining at the expense of less energy-intensive ones" (Sorrell and Dimitropoulos, 2006). This effect is considered by some authors to be the main cause of the constant growing trend in energy consumptions (Ayres, van den Bergh, 2005; Sorrell and Dimitropoulos, 2007). Nevertheless, the methodology and the extent of the impact on the economy energy path remain unanswered questions (Dimitropoulos, 2007). Here, the macro rebound effect is analyzed in the context of the productive structure's shape, as a result of the energy efficiency in the freight transport sector. This work will not tackle the question of the rebound size, nor the dispute over the proper econometric/theoretical approach. It will propose instead a novel view on the issue in order to create a bridge between different disciplines and deal with the broader question of the relationship between energy level, energy efficiency, and complexity.1 Although the analysis is qualitative, it provides a link between 1 We here intended for energy efficiency to mean an energy conversion process that transforms energy (chemical or thermal) into work and for complexity to mean a feature of the whole structure of the system. Two hierarchical levels are henceforth introduced: the process (the conversion of energy) and the system. A new, more efficient process with a higher energy conversion rate may not lead to a lower energy level for the system, given the same number of components and extent of the system. If the system modifies its structure in a way that raises the number or length of interactions between components per unit of time (intensity) or in other words, if the system alters its complexity level, then a higher energy level will be achieved. 527 the dynamics of production patterns and the effect of efficiency increases, in the light of the macro-economic effects of increased energy demand. 2. The rebound effect in the freight transport system In the aftermath of the first Oil Crisis (in the year of 1973), developed countries underwent a remodeling process in the fields of energy final use and energy production. Both in the EU and in the U.S., the road freight transport systems were subject to major changes. Some measures taken include: engines technology enhancement, aerodynamics, size and speed limits and market deregulation.2 A significant change in fuel economy occurred thereafter in long-distance shipments in the following decade. However, energy efficiency improvements were initially employed to reduce fuel economy, but later to reduce both fuel consumption and improve vehicle performance. 2.1. Adjusted fuel economy: an experimental basis for energy efficiency assessment Unfortunately, a reliable energy efficiency database for the EU freight transport system fails to exist. Published related figures have mostly been derived from manipulations of aggregate data, such as the domestic consumption of fuel and the overall distance traveled by carriers. Such estimations, however, were inevitably affected by the driving conditions, load and size factors and available power of engines. To account for these influences on energy efficiency data, we measured the evolution of an adjusted fuel economy of a sample of 100 trucks and 16 different European brands, between 1978 and 1995, grouped by size, on the basis of road tests taken in the same standard condition and at the same average speed (Tuttotrasporti, 1978– 2005). The adjusted fuel economy is the fuel economy value recorded on an even road at 80 km/h and divided by the weight of the vehicle (full load) and the maximum power of the engine. It is thus measured in liters per kilometer and multiplied by tons and HP. The data were organized into four size groups and the adjusted fuel economy was measured as a mean of small samples of vehicles (Table 1). 2 Turbocharged diesel engines were introduced in the early 1980’s and traction control systems were also developed during the same time period to optimize fuel economy. Inter-cooling systems and aerodynamic advancements (cab air deflectors) were developed in the mid 1980's, and further improvements were made to tires (low profile radial tires and multiple trailers). In many European countries in the early 1970's, new laws raised speed limits and maximum load size. This trend has been maintained in the following years, reaching a maximum speed between 80 and 90 km/h and a maximum truck weight of 40 tons for international transport. The Motor Carrier Act of 1980 reduced restrictions on entry and expansion in the trucking industry and relaxed various regulations. The Surface Transportation Assistance Act (1982) modified state requirements on size and weight limits for trucks (DOE/EIA 1995). On May 22, 1985, the European Court ruled in favour of the European Parliament. In the following years, the European Council submitted many directives that were in favor of transport market liberalization and integration (OECD, 1997). 528 EC O LO GIC A L E CO N O M ICS 6 7 ( 2 00 8 ) 5 2 6 –5 37 Table 1 – Fuel Economy and Adjusted Fuel Economy Vehicle size Number of trucks evaluated b 3.5 ton 3.5–5 ton 8–18 ton N 43 ton 20 21 19 51 Fuel economy: 1978–1990 Fuel economy: 1978–1998 Adjusted f.e.: 1978–1990 Adjusted f.e.: 1978–1998 −16% −17% −14% −21% −36% −25% −16% −27% −21% −38% −33% −37% − 58% − 46% − 38% − 50% Source: Tuttotrasporti. All estimations converge towards a 30% reduction between 1980 and 1990. An improvement in energy efficiency of about 30% is remarkable. Nonetheless, it still underestimates the real efficiency leap that increased economy during the last three decades. None of the abovementioned statistics on freights account for the efficiency improvements conveyed by the steady shift towards heavier loads. Thus, a bigger truck has a lower specific consumption per unit of gross weight than a lighter one.3 In order to draw a definitive assessment of the overall energy efficiency improvement, we must include the effect of the “size factor” when considering the adjusted fuel economy. The size factors depend on the evolution of weight limits and the composition of the vehicle fleet. Both aspects indeed have consequences on the average weight of trucks and thus on their fuel economies. The first factor is difficult to estimate because of the diversity of national legislations. The second factor affecting the aggregate fuel economy of trucks is the composition of the vehicle fleet. In Europe, the balance between single unit trucks and combination trucks shifted by 33% toward the combination trucks.4 The latter actually increased in 25 years by 282% and the former only by 165% (EUROSTAT, 2005). Nevertheless, in the absence of a sufficient database concerning both size limits and truck fleet composition, we can still draw some conclusions based on the fragmented resources currently available (Tuttotrasporti, 1978–2005; CONFETRA, 1998). If we reasonably assume that in 1980, medium size trucks weighed, on average, 12 tons and trailers weighed about 30 tons, the shift would have caused an efficiency improvement of about 7% in the fuel economy of the whole fleet. In addition, since 10% of the fleet belonged to the last category, we should add a 1.5% efficiency improvement due to the increase in size limit from 32 to 44 tons. We may thus estimate that energy efficiency improvements due to changes in size ranged from 8% to 9% in EU. Ultimately, the veritable energy efficiency improvement in the road freight transport sector during the two decades from the second half of the 1970's to the first half of the 1990's was about 40%. 2.2. Trends in the transport system and economic indicators Despite the drop in specific fuel consumption of trucks, energy consumption in the freight transport sector increased at a rate 3 The size-effect is mainly caused by aerodynamics because about 30% of energy loss is due to air resistance. 4 Combination trucks are normally twice the size of single unit trucks and are mainly devoted to international transport. unmatched by any other sector. Around the world, energy final use in the industry sector increased between 1970 and 1995 by 45% and in all others sectors between 51% and 58%, while it nearly doubled, growing by 90% in the transport sector (IEA, 1997; EUROSTAT, 2005). In the transport sector, road mode showed the highest growth rates. Similarly, the fuel use increased between 75% and 100%, with an average annual growth rate of 3–4%. In Europe, the related statistics are even more striking. During the same period in the EU, tonne-kilometers of goods hauled by road increased by 130%, while the freight transport sector generally grew by 65%. However, if we compare trends in the freight transport sector to some economic indicators like the real GDP and the industrial production, especially during the first decade between 1970 and 1995, it appears impossible to strictly link the growth in tkm to the fundamentals of economy. In the EU, tkm showed growth rates much greater than both the GDP and industrial production (Table 2). 2.3. Globalization and outsourcing Globalization was indeed predominately the driving force behind most of the changes in the economy that occurred between the 1970's and 1990's. It reshaped most economic sectors and also deeply affected society. However, if globalization was the process promoting evolution in the economy, the transport sector must have been the medium responsible for developing its changing power. For the sake of this research, we can consider globalization to be the composition of two main processes undertaking the productive sector: 1. The integration of markets (any markets, goods, finances, production means and materials) 2. The shift from a fordian to a postfordian mode of production We describe a fordian system of production as a model where the productive chain is mainly set in a unique plant or place, to and from which materials and products depart. On the contrary, in a postfordian system, the productive chain is scattered throughout the territory, often over different countries (Lipietz, 1992; Kitschelt et al., 1999). This reshaping of the productive chain mirrors, in physical terms, a different structure of costs and rights that sprang from a new way of sourcing by firms. This new form of providing services, materials or productive means from outside is what is called outsourcing. As stated in an OECD re port: “Globalization of industry refers to the operations of firms undertaken to organize their development, production, sourcing, marketing and financing activities. A distinctive feature of the globalization phenomenon is the 529 EC O L O G IC A L E C O N O M IC S 6 7 ( 2 0 08 ) 52 6 –5 37 Table 2 – Freight Transport: EU-15 (4 modes) Road Rail Inland waterways Sea (domestic) Total, 4 modes Real GDP Industrial Production 489 720 976 1.010 1.124 +47% +56% +130 % 282 290 255 234 222 + 3% − 23% − 22 % 103 106 107 107 115 + 3% + 8% + 12% 472 781 923 955 1.070 +65% +37% +126% 1.346 1.897 2.262 2.307 2.531 +40% +33% +88 % + 3,6% + 2% + 2% + 1,9% + 2,3% + 30% + 29% + 63% – 0% +2,1% −1,4% +3,7% +18% +23% +56% 1000 millions of tonne-km 1970 1980 1990 1991 1995 1970–80 1980–95 1970–95 Source: EUROSTAT, OECD. division of firm operations into separate segments carried out in different countries.” (OECD, 1996). Outsourcing can thus be considered a distinctive feature of the postfordian model of production. On the one hand, it caused a significant reduction in production costs (especially those related to the storage of goods and productive means) through an improvement in activities specialization (cluster of small firms), an optimization of management costs (downsizing) and a reduction in labour costs (obtained with the exploitation of new labour markets). On the other hand, outsourcing dramatically increased both the distance of travels and the intensity of trades.5 It relied mainly on lower transportation costs, which permitted an expansion in markets’ physical borders, an integration of markets once isolated and the placement of segments of production in different places. In other words, outsourcing transformed trucks into the new warehouses of firms. Globalization has thus most likely affected the factors market more deeply than the products market. 2.4. The causal relationship: an efficiency improvement prior to globalization It is now possible to question whether any causal relationship between fuel economy energy efficiency in the freight transport sector and the formerly mentioned change in the productive structure can be established. During the 1970’s, in Europe, fuel costs amounted to around 30% of total costs in the long distance road freight transport sector (Tuttotrasporti, 1978–2005). If we thus consider an overall efficiency improvement of 40%, shipping costs would drop by 12%, which is a remarkable reduction. The efficiency leap would 5 “International sourcing of parts and materials is a major feature of global production systems and sourcing accounts for a large part of total trade. There is some evidence that international sourcing is linked to international investment. Industries that have high levels of international inward investment are also more likely to source internationally. OECD work which examined international sourcing in the manufacturing industries of six large countries (..) points to the interdependence of their economies. The study concluded that direct imports of manufactured intermediate inputs from abroad (which range from 50 to 70% of manufactured imports for the six countries) rose more rapidly than domestic sourcing in all countries, with the highest growth rates occurring more recently” (OECD, 1996). have affected transport costs in such a proportion to suggest that the productive structure was also affected. For example, efficiency raised the relative costs of storing materials and products that, in the long run, would foster firms to resort to outsourcing more than internal production. Major changes in truck fuel economy occurred in Europe in the decade preceding 1980. These changes were in fact triggered by the first oil crisis of 1973. The start of globalization is commonly believed to be in the following decade mainly because of the dramatic increase in foreign direct investment by international firms (OECD, 1992, 1993, 1996, 2002; World Bank, 2005). In order to set a defined time for the “dawn” of globalization – a phenomenon that predominately involved international outsourcing in the productive sector – a guiding indicator must be discovered among the various possibilities. The best measure of international outsourcing is therefore the intra-industry rate on trades. The intra-industry indicator is the rate of trade within the same industries over the entire trade of goods between countries. This indicator thus measures the percentage of parts, components, raw materials, and productive means over the amount of products exchanged. According to this indicator, intra-industry trade grew by 36% from 1970 to 2000 in the OECD countries. For example, the percentage of parts and components over the total of exports grew from 2% of the total to 10% in Europe and from 6% to 15% in USA and Japan between 1980 and 2000 (World Bank, 2005). The intra-industry indicator indeed displayed a dramatic leap such as to overshadow a change in the productive systems. It increased more than any other related figures, such as the percentage of Export and Import over the GDP, the intra-firm indicator (OECD, 1993). Nonetheless, according to the same figures, such a leap occurred in the second decade of the range of time considered (Table 3). It was indeed between 1980 and 1990 when this indicator further grew in all the OECD countries and in Europe (Table 3). These figures suggest that the main transformation in economic structure affected international trade after 1980 and therefore, following the abovementioned change in truck fuel efficiency. Although a temporal sequence does not entail causality between events, it does provide concrete evidence that the preceding event cannot be caused by the subsequent one. The dramatic change in truck fuel economy occurred before the rise of international outsourcing as a distinctive feature of globalization. The leap in fuel efficiency was not caused nor driven by international outsourcing or globalization, but it is 530 EC O LO GIC A L E CO N O M ICS 6 7 ( 2 00 8 ) 5 2 6 –5 37 Table 3 – Intra-industry indicator (share on international trade): growth rates and some annual values 1970–80 1980–90 1990–2000 1970 1995 United Kingdom France United States Japan Italy Germany EU (14) OECD (23) 39% 4% 4% − 20% 13% 1% 10% 6% 13% 10% 55% 89% 5% 28% 20% 32% 3.6% 2.7% 5% 10% 3% 5% 14% 4% 53% 67% 44% 21% 49% 56% 56% 49% 73% 77% 65% 41% 64% 72% 71% 67% Source: OECD. very likely that it was a driving force, among others, of those events. 3. The traffic density growth: Italy case study The shift from a uni-located (fordian) productive chain to a pluri-located (postfordian) one, together with the integration of markets, placed an increasing burden on the road transport system. The fuel use in the freight transport sector grew not only because the mean distance of travels augmented, but also because the frequency changed. In other words, the outsourcing system of production is strictly connected with a more intensive and flexible transport system, which was provided solely by the road modal sector. According to this thinking, the growth in tkm is not only due to market integration, but also to the outsourcing process. Did the traffic density grow along with the tkm? According to OICA (Organisation Internationale Constructeurs de Automobiles), traffic density of goods vehicles grew more than that of private cars in Europe. Between 1980 and 2000, truck density almost doubled in Europe while car density grew much less. Nevertheless, this approximation fails to be a sound indicator of how much the traffic density of trucks grew locally, the desired value in this case.6 These figures actually match the number of registered vehicles with the length of the road network (Table 4). In order to get a realistic picture of local traffic density, it is important to reduce the scope of the analysis. A case study has consequently been taken into account. For this analysis, it is necessary for traffic density data to be consistent over time and space and thus, a single country has been considered. Results of this analysis can be considered reliable, but we would recommend an extension of the analysis to a greater number of countries. We analyzed data from the traffic census in Italy between 1972 and 1989 (ANAS, 1973–1990). The traffic census in Italy roughly occurred every five years and was set on 398 tracking points scattered throughout extra-urban roads all 6 It is important to detect traffic density on the local road network because as the frequency and not only length of shipments grow, the secondary road network is also affected. On the contrary, if only the distance of shipments grow, as a consequence of a mere integration of markets, the highway road network is predominately impacted. over the Italian road network. These sites have remained relatively unchanged over the years. Data were collected on eight days (six working days) and four nights during all seasons. Vehicles were arranged into eight categories, four of which were goods vehicles. DAT (Daily Average Traffic) expresses a weighted average7 of the vehicles on the road every day during the year. We selected a sample of 44 tracking points scattered all over Italy that were significant based on location (on roads connecting either productive sites or big urban conglomerates) and traffic density. According to these data, while total vehicles grew by 25% in 25 years, goods vehicles (over three tons) grew by 132% (much more than the 54% growth in industrial production during the same interval of time). Semi-trailers trucks, used mainly for international duties, grew by 172% (Table 5). This statistic matches the growth rate in tkm (180%) that occurred in Italy during the same time period, further confirming the impression that shipments increased much more for the density than for the distance of travels. It is important to demonstrate that freight transportation grew not just in distance traveled, but also in frequency of shipments. The density of connections can be considered as a footprint of structural complexity of the productive system and the reason why will be addressed in the next paragraph. These data seem to confirm the hypothesis that both distance and density of movements grew in the two decades analyzed here and that integration of markets coupled with structural change in the production system should be regarded as the driving factor of such a growth. 4. The complexity leap The underlying hypothesis of this work is that higher complexity counterbalances, on a global scale, the effects of higher efficiency on a process scale. So far, we have focused on demonstrating the actual extent of efficiency improvements and the timing with respect to the evolution of the productive structure. However, the fact that the productive structure, all through globalization, evolves towards higher complexity was taken for granted. We now want to stress the extent to which the economic productive system increased in complexity during the last two decades.8 It is beyond the goals of this analysis to ascertain what complexity is or how it should be approached. The scientific community has been unable to establish or agree on a universal definition or paradigm of complexity and any attempt to univocally measure complexity is therefore doomed to failure. In the present analysis, we will “identify complexity in two operational ways: as a measure of the information needed to describe a system's structure and function, or as a measure of 7 The average is weighted over the course of one year in order to represent the daily traffic of an average day and thus, to balance seasonal peaks and tufts. 8 In his work, Ayres stresses how, in order to understand the energy path of developed countries after 1975, it is important to introduce a “structural shift” in the productive system. He maintains that energy and GDP growth can be explained as a result of the energy efficiency improvement up to that date, but a structural change ought to be taken into account thereafter (Ayres and Warr, 2005; Ayres, van den Bergh, 2005). 531 EC O L O G IC A L E C O N O M IC S 6 7 ( 2 0 08 ) 52 6 –5 37 Table 4 – Traffic Density in EU and Italy Vehicles per km of Road 1980 1990 2000 1980–00 Europe, CV’s Europe, Cars Italy, CV’s Italy, Cars 3.6 33.9 4.7 60.2 5.6 42.6 8.0 90.2 6.7 50.1 9.1 100 86% 47% 93% 66% Source: OICA, Europe = 15 countries, CV = commercial vehicles. the rate of energy flowing through a system of given mass” (Chaisson, 2001). It will thus be assumed that a more complex system consumes more (per unit of mass and time) and the complexity we refer to is a structural or morphological complexity as we are dealing with systems with undefined boundaries and innumerable components, such as the productive and transport systems. The two main assumptions regarding complexity that we are concerned with are: • A more complex system consumes more energy per unit of mass and unit of time (higher energy density rate). • Structural complexity primarily concerns the components' organization9 rather than the components' variety or number. Further remarks capture the duality efficiency/complexity. We should bear in mind that while we are referring to energy efficiency improvements, we are dealing with a process-scale analysis, whilst the leap in complexity concerns the globalscale analysis. These phenomena are at two different hierarchical levels: • Energy efficiency concerns energy converting processes and is therefore at the components level of the system. • Complexity (structural) concerns the organization of the system and is thus at the global level of the system. In order to shed new light on the complexity change that occurred in the productive system and in order to establish whether such a change must be considered augmentative, we now focus on the two major aspects of globalization. We will tackle the goal of detecting complexity change through several points of view, for the concept of complexity is not, as already asserted, unique. We will dissect the globalization process and examine each piece through various complexity theories in order to establish a clear trend in changes. Each complexity theory will therefore be used as a tool or better stated, a paradigm, on the basis of which we can make a decision about the direction of system evolution. The two main features of globalization to be addressed are: zation process, is due to the abolishment or the relaxation of commercial barriers and is affected by both goods (final consumption) and factors markets. Before globalization, markets were mainly nationwide, but after globalization, the scope of trade extended to nearly the entire world. Barriers to market integration are physical (shipment costs) or economical (taxes). It is important to point out that both flaws curb the rise of global free market. It would be insufficient to remove custom borders if transport technology failed to reduce the costs of shipments (Krugman et al., 1995). Technological and economical aspects together participate in the merge of previously separate markets. We will focus on the technological aspects which constitute premises or consequences, according to different views on the subject, to free trade economic policies. It is of little relevance whether the former comes before the latter, or vice versa for the present analysis. It is indeed true that energy efficiency improvements in the transport sector play major roles in market integration. If energy efficiency led to market integration, then to what extent did the productive system consequently grow in complexity? There are at least two theories which may assist us in defining a complexity trend in the present case: one comes from ecology and the other from economics. In fact, we argue that complexity grew because of: 1. An increase in hierarchical levels of the system 2. The scale economy effect or the idea that more specialization of economies leads to more integration and interdependence of the system 4.1. According to hierarchical theory of ecosystems (Odum 1996, Allen and Starr, 1982), complexity increases when a new, higher control level is introduced within system components. According to this view, complexity grows when new parts are added in such a fashion to “exert new constrains over the others.” Otherwise, the system grows just in “complexification” (Tainter et al., 2003). That is to say, a system is more complex when, by adding parts, it increases the number of kinds of relationships. In contrast, a system becomes more complicated when adding parts causes a mere growth in the number of relationships. Key concepts of hierarchical system theory are: the breakdown of symmetry and reduced degree of freedom for lower level variables. These two features account for the “growth Table 5 – Italy, DAT (daily average traffic) in Extra Urban Road Network Goods Single unit Trailers SemiTotal vehicles trucks over trucks trailers vehicles up to 3 ton 3 ton trucks 1. The integration of markets (factor and goods) 2. The shift from a fordian to a postfordian productive system We will begin by analyzing the process of market integration. Market integration, which is at the core of the globali- 9 For organizational purposes, we refer to any system’s components acting or arranged in a cooperative, systematic fashion. Market integration, first paradigm: hierarchical theory 1972 1979 1985 1989 673 669 881 840 24.75% 624 602 639 566 −9.25% 335 296 261 239 − 28.80% 112 171 246 301 169.56% Source: elaboration on ANAS traffic census, 1973–1990. 12.297 11.451 14.199 15.340 24.75% 532 EC O LO GIC A L E CO N O M ICS 6 7 ( 2 00 8 ) 5 2 6 –5 37 Fig. 1 – Market integration and hierarchical system theory: symmetry breakdown. in organization” of the system. The two graphs (Fig. 1) reveal that the system grew in number of components and connections. Only the second graph, however, illustrates a growth in organization. Symmetry is broken in the second case and thus the information that flows from A to B (or from B to A) is the same, despite the number of “B,” while the information that goes from 2 to 3 is different from the information flowing from 2 to 1. If you add B to A, you preserve symmetry, but if you add 3 to 2, you create asymmetry in the system. The same is true for the particles' degree of freedom. If you add a new B to the system, the degree of freedom of former particles does not change. If you, however, introduce a new control level, such as 3, the degree of freedom of both 2 and 1 is reduced. For example, suppose you want to set up a European championship. According to the theory, the degree of freedom of the top teams participating in the national championships is now reduced because the teams must avoid being relegated as opposed to simply compete for first place. They are therefore competitors instead of champions. If you add more teams to the groups, you will consequently have more teams running for the national title. Champions will therefore have to play more teams, but will retain their title. This is, more or less, what happened to firms after globalization merged national markets. Firms that were big and strong enough to access the “world championship” had to extend and reorganize their network of production and logistics. International corporations have districts (control nodes) at any territorial scale (local, national, transnational). How does this change affect the movement of goods? Suppose some national supermarket chains merge to form a big international company. Management at this higher level must have contracts with a unique international supplier instead of several local ones. Goods now arrive at all warehouses from the same supplier, regardless of the nation or the distance of provenience. The same can be said for factors in the productive system. Market integration nonetheless changed big companies’ logistics. 4.2. Market integration, second paradigm: geographical gradient Capital intensive sectors, those sectors with a higher burden of structures and machineries over production than manufacturers, dislike scale economies (Milgrom and Roberts, 1992). For such sectors, costs are reduced when production is significantly increased because structural costs are higher than variable costs. Due to economic scale effects, firms tend to increase in size or productive districts to form. Economic scale effects were fostered by globalization and market integration because more competitive firms could access new, large markets. Economic scale effects brought the new factor of selection to firms and thus survivors grew and engulfed or displaced minor firms. One effect on national economies was the growth of strong national industries and the disappearance of weak ones. In the long run, national economies tend to be specialized. The increased specialization of economies can be regarded as a factor of complexity growth. National specialization can be seen as a geographical gradient in the productive system.10 This gradient leads to a breakdown in symmetry, with the rise of a new hierarchical organization level, and to increased average distances of parts (production plants). It will therefore augment both the distance and the frequency of shipments, as already remarked in light of the hierarchical system theory. An example is shown in the graph of three countries and three industries (Fig. 2). At the beginning, there is no geographical gradient, but with specialization, similar industries tend to set themselves apart. As a result, there are three countries, each with one industry. Like organs in a body, they now need to become coordinated at a 10 It is a spatial gradient that relates the diversity of the system to its size and it is an ecological function. In ecology, in fact, biodiversity it is also related to the size of the ecosystem (Fath et al., 2001). EC O L O G IC A L E C O N O M IC S 6 7 ( 2 0 08 ) 52 6 –5 37 533 Fig. 2 – Market integration and geographical gradient: the economic scale effect. higher control level. It is also very simple to show that the average distance among components thereafter increased. Complexity, therefore, increased. 4.3. Productive system shift, third paradigm: homeostatic theory When analyzing the structural complexity change resulting from globalization, of paramount importance is the shift from a uni-located, national productive chain to one that is plurilocated and international. This process is a complexity leap because it brought about: • More specialization and more integration (interdependence) among system components and suppliers and subsidiaries became essential (synchronization) • An increase in interactions among system components due to changes in the shape of the production chain (which will be addressed in the next section) The first feature mainly concerns the outsourcing process because it affects firms by economically externalizing their functions. For those firms relying on external resources to pursue their productive needs, production becomes less costly, but more subdued due to uncontrollable factors. Part of its activity, formerly controlled managerially and internally, is now focused on free market. This shift reduces the stability of the system and increases its complexity. The homeostatic system theory provides us with a clear explanation of the kind of change that occurred in productive systems, thus prompting economic externalization (Odum, 1997). The first graph (Fig. 3) represents production regulated by a feedback loop (El-Samad et al., 2006). The stock of goods in the warehouse is determined by two fluxes, sales and production volume. In the first, simple example, management determined production according to one feedback indicator: the amount of products stocked (which is proportional to sales). When the stock decreases below a certain level (set point), the production receives a positive input (stimulator message). Suppose we now want to regulate the production in accordance with our sales and profits. We will therefore have two controls (we increase the production when sales or profits increase) and one feedback loop (sales and profits are proportional and cannot be counteractive). This is a strengthened version of the previous loop based solely on stimulators’ messages. This kind of homeostatic system works exactly like thermostats and are quick in positive feedbacks but slow in negative ones. It is otherwise obvious that such systems tend to grow rapidly in terms of production but are much less capable of diminishing it. If we add an inhibitor factor (negative control) to the stimulator factor (positive control), we create an inner regulatory loop which increases both the complexity and the adaptability of the system. Suppose we now buy part of the production on the market. Previously, we stocked all the materials and adjusted the production according to sales, but now we have to take into account both revenues and costs. We therefore introduce a negative feedback control into our loop (costs of parts). In this case, the loop, from a homeostatic perspective, increases in complexity (Fig. 3). This phenomenon involved outsourcing because firms dealt with costs (due to bargain, timing and also delivery) more than ever. How can this process lead to an overall increase in the movement of goods? The homeostatic approach provides a sound explanation for why and how complexity grew when the productive system shifted towards outsourcing (El-Samad et al., 2006). According to this view, the amount of information flowing through the system increased. However, it is much less clear whether this information is mainly monetary or whether it also has some content matter. Did the flow of products, semi-products, and raw materials increase in response to such a complexity leap? From this analysis, it is impossible to establish any clear tendency and further research is necessary. 4.4. Productive system shift, fourth paradigm: graphic theory The last part of this analysis and by far, the most crucial, describes the shape of the productive structure. It was already made clear that the main feature of the shift from a fordian to a postfordian system concerns the location of the productive chain. Formerly, the productive chain was set thoroughly in one plant, to which raw materials were delivered and from which products were shipped. From the 1970's onward, big companies began disassembling the production chain and placing it in several, scattered structures, often in foreign countries. We will use graph theory in an attempt to illustrate how the productive structure evolved in complexity. In order to apply 534 EC O LO GIC A L E CO N O M ICS 6 7 ( 2 00 8 ) 5 2 6 –5 37 Fig. 3 – Productive system shift and the homeostatic paradigm: decentralization of controlling. this tool to the economic structure, the system has been drawn to define and portray its shape. The graphs sketched are intended to be approximate, but not exact representations. Graph theory, however, helps to establish whether a determined shape can be considered more complex than another. It is therefore suitable to compare the complexity of abstract structures, but cannot be considered a tool for measuring complexity. Graphs A and B (Fig. 4) display some sketches of the productive structure. They solely focus on the inflow of materials, the supply and productive chain, and assume that the outflow, the product distribution chain, is not affected by the structural change being analyzed. Graph A shows a “fordian” factory in which the materials of semi-products flow from suppliers to the center where the productive chain is concentrated (the star-like graph). Graph B (Fig. 4) shows a “postfordian” factory in which the suppliers become part of the “spread” productive linear chain (the square-like graph). The dashed line indicates that we can consider it to be both an open loop (no main plants) and a closed loop if the final assemblage occurs in the plant where the process started. Graph B is more complex than graph A even though it has the same number of points (and lines, in the open loop case) because: I. The minimum degree (minimum number of incident lines to each point) of graph points is higher in the closed loop case. II. Connectivity (minimum number of points whose removal results in a disconnected or trivial graph) is higher in the closed loop case. III. The path length is four (three if open loop) times that of graph A. IV. The number of possible path-cycles across the points is two times that of graph A. In other words, there are, at least, two non-isomorphic graphs (graphs B and C) that connect the points in a chain, while there is only one in a star-like structure like graph A.11 V. The number of non-isomorphic graphs increases if we consider a chain of more than four points (graphs 1 to 4), while the number of star-like graphs remains the same. This means that there is just one structure that connects many points to a center while there are many different 11 In other words, there is just one way to go from the periphery to the center, regardless of the number of nodes considered, while there are many ways to connect the same number of points in the path. Furthermore, the number of different ways increases with the number of points. This does not mean that, in a scattered productive chain, factories (points), are connected randomly, but instead that there are multiple ways for a chain to develop its pattern and just one for a centralized system. EC O L O G IC A L E C O N O M IC S 6 7 ( 2 0 08 ) 52 6 –5 37 535 Fig. 4 – Productive system shift and graph theory: the shape of the productive chain. structures that make a chain (open or closed) out of the same number of points. All of the above statements suggest that the second structure (postfordian) presents a higher degree of freedom (an increase in multiplicity of the system) and thus relates to a more complex system. If we generalize this approach to the entire productive system, we draw two different pictures: one system resembles a sky of scattered stars and the other, a web. In fact, each chain opens branches to side-suppliers, which, due to increased specialization, are connected to other productive chains (Fig. 5). The web-like structure ultimately increases the degree of freedom of the system because it presents an even higher minimum degree of graph points, more point and line connectivities (which are unchanged in the star-like structure), and longer graph paths. Although the fordian structure depicted in Figs. 4 and 5 may seem overly simplified, as it is reasonable to assume that connections occur also among suppliers or from suppliers to various factories in such a way that the first picture would also resemble, to a lesser degree, a net, the conclusion regarding the degree of freedom of the two systems remains the same. In the second structure, the degree of freedom is higher. A system with a higher degree of freedom is a more complex system in the sense that, as for any physical system, it has increased multiplicity or number of different available states. In other words, a more complex system has more ways to dispose of particles, in this case, goods or raw materials, and therefore to dissipate energy. 5. Conclusions It has been shown by several qualitative approaches that the complexity of the productive system has grown in the last three decades. According to all but one (Fig. 3) of these approaches, complexity growth also caused the major 536 EC O LO GIC A L E CO N O M ICS 6 7 ( 2 00 8 ) 5 2 6 –5 37 Fig. 5 – Productive system shift and graph theory, bis: generalization to the whole productive system. circulation of materials, in terms of distance and frequency. Such a complexity leap was augmentative of the number, in time, of relationships among its parts and explained the growth in energy density rate thereafter. Figures (Table 4 and 5) reveal that traffic density growth occurred during the interval of time here considered and reflect the transformation of the productive system in the direction of increasing the number of relationships within its components (and not solely the length). This transformation is therefore consistent with the hypothesis so far investigated. It is here suggested that the complexity leap was, under these circumstances, the strategy for the system to dispose of the increased energy made available by the efficiency change. Such a strategy, however, does not arise from nothing. It represents a particular disposition and organization of system components previously available, yet not economical. When boundary conditions change such that more economical paths are no longer exploitable and if there is an energy inflow such that a constant pressure exists upon the system, the system will thereby explore new settings in order to dissipate energy flow. In the words of Chaisson: “The phenomenon is also sometimes called ‘self-organization,’ although that term and others like it (those with the prefix ‘self-’) are deceptive in that such ordering is actually occurring not by itself, as though by magic, but only with the introduction of energy” (Chaisson, 2001). Therefore, in our opinion, we should approach the question of the rebound effect, in light of the dynamic relationship between energy efficiency of the process and complexity level of the system. Such a relationship concerns both biological and economical systems. In nature, more efficient processes correlate to more complex systems (and energy costs). This analogy probably underlies a more general thermodynamic mechanism common to many evolutionary systems. Such a mechanism boils down to the geometrical features of the system and the way it molds itself as to host the larger energy influx. In the words of the great biologist and pioneer of evolutionary thermodynamics, Alfred Lotka: “The physical geometry and environmental conditions select the path for the energy flow; if the crack under the door is large, it will rapidly dominate the other two paths, and the out flowing energy will soon return the room to equilibrium with its surrounding” (Lotka, 1956). Acknowledgments We would like to thank Professor Sergio Ulgiati and Professor Eric Weeks for their comments and advice. We are also very grateful to Farrah Elchahal for her kind and meticulous revision of the manuscript. REFERENCES Alberts, B., Bray, D., Lewis, J., Raff, M., Roberts, K., Watson, J.D., 1994. Molecular Biology of the Cell. Garland Publishing, Inc., New York. Allen, T.H.F., Starr, T.B., 1982. Hierarchy: perspectives for ecological complexity. University of Chicago Press, Chicago, IL. ANAS, 1973–1989. Censimento della Circolazione Lungo le Strade Statali ed Autostrade, A.N.A.S., Direzione Centrale Tecnica. Ayres, R., van den Bergh, J., 2005. A theory of economic growth with material/energy resources and dematerialization: Interaction of three growth mechanisms. Ecological Economics 55, 96–118. Ayres, R., Warr, B., 2005. Accounting for growth: the role of physical work. Structural Change and Economic Dynamics 16, 181–209. Berkhout, P., Muskens, J., Velthuijsen, J., 2000. Defining the rebound effect. Energy Polici 28, 425–432. Brookes, L., 1990. Energy efficiency and economic fallacies. Energy Policy March 783–785. Chaisson, E., 2001. Cosmic Evolution — The Rise of Complexity in Nature. Harvard University Press, Cambridge, Massachusetts, London, U.K. CONFETRA, 1998. Effetto Serra, Emissioni di CO2, Trasporto Merci, Quaderno N. 109/1 — Novembre 1998, Centro Studi Confetra. Dimitropoulos, J., 2007. Energy productivity improvements and the rebound effect: An overview of the state of knowledge. Energy Policy 35, 6354–6363. DOE/EIA, 1995. Measuring Energy Efficiency in the United States' Economy: A Beginning Distribution, Category UC-950, Energy Consumption Series. El-Samad, H., Prajna, S., Papachristodoulou, A., Doyle, J., Khammash, M., 2006. Advanced methods and algorithms for biological networks analysis. Proceedings of the Ieee 94 (4), 832 Postprint available free at: http://repositories.cdlib.org/ postprints/1474. EC O L O G IC A L E C O N O M IC S 6 7 ( 2 0 08 ) 52 6 –5 37 EUROSTAT, 2005. Energy & transport in figures 2004, European Commission, Directorate General for Energy and Transport. Fath, B., Patten, B., Choi, J., 2001. Complementary of ecological goal functions. J. Theor. Biol. 208, 493–506. Geening,, L., Greene,, D., Difiglio, Carmen, 2000. Energy efficiency and consumption – the rebound effect – a survey. Energy Policy 28 (6/7), 389–401. IEA, 1997. Indicators of energy use and efficiency. IEA, Paris. Khazzoom, J.D., 1980. Economic implications of mandated efficiency in standards for houshold appliances. The Energy Journal 1, 21–40 No.4. Kitschelt, H., Lange, P., Marks, G., Stephens, J.D., 1999. Continuity and Change in Conteporary Capitalism. Cambridge University Press, Cambridge U.K. Krugman, P., Cooper, R.N., Srinivasan, T.N., 1995. Growing world trade: causes and consequences. Brookings Papers on Economic Activity, Vol 1995, No. 1, 25th Anniversary Issue, pp. 327–377. Lipietz, A., 1992. Towards a New Economic Order, Posfordism, Ecology and Democracy. Polity Press, Cambridge U.K. Lotka, A., 1956. Elements of Mathematical Biology. Dover Publications, Inc, New York. Milgrom, P., Roberts, J., 1992. Economics, organization and management. Prentice-Hall Englewood Cliffs, NJ. Odum, E.P., 1997. Ecology: A Bridge Between Science and Society. Sinauer Associates, Inc., Publishers, Sunderland, Massachusetts 01375 U.S.A. Odum, H.T., 1996. Environmental Accounting. Emergy and Environmental Decision Making. John Wiley & Sons, Inc., New York, USA. View publication stats 537 OECD, 2002. Intra-industry and intra-firm trade and the internationalisation of production. Economic Outlook, vol. 71. OECD, Paris. Part VI. OECD, 1992. Economic Regionalisation and Intra-Industry Trade: Pacific-Asian Perspectives, OECD DEVELOPMENT CENTRE, Working Paper No. 53 OECD, 1993. Globalisation and Intra-Firm Trade: an empirical note. OECD Economic Studies, Paris. No. 20. OECD, 1996. Globalization of Industry, overview and sector reports. OECD, Paris. OECD, 1997. Liberalisation and Structural Reform in The Freight Transport Sector in Europe. OECD, Paris. Saunders, H., 1992. The Khazzoom–Brookes postulate and neoclassical growth. The Energy Journal 13 (4), 131. Sorrell S., Dimitropoulos J., 2006. The rebound effect: micro-economic definitions, limitations and extensions. UKERC Working Papers, SPRU, University of Sussex. Sorrell, S., Dimitropoulos, S., 2007. The rebound effect: microeconomic definitions, limitations and extensions. Ecological Economics, doi:10.1016/j.ecolecon. 2007.08.013. Tainter, J.A., Allen, T.F.H., Little, A., Hoekstra, T.W., 2003. Resource transitions and energy gain: contexts of organization. Conservation Ecology 7 (3), 4 [online] URL: http://www.consecol. org/vol7/iss3/art4. Tuttotrasporti, 1978–2005. Editoriale Domus S.p.A., Via Gianni Mazzocchi 1/3, 20089 Rozzano (MI). World Bank, 2005. Global Economic Prospects 2005. Washington, D.C.