Constructing physical input output tables for environmental modeling and accounting: Framework and illustrations

ECOLOGICAL ECONOMICS 59 (2006) 375 393 available at www.sciencedirect.com www.elsevier.com/locate/ecolecon ANALYSIS Constructing physical input output tables for environmental modeling and accounting: Framework and illustrations Rutger Hoekstra a,, Jeroen C.J.M. van den Bergh b a Statistics Netherlands, Prinses Beatrixlaan 428, 2273 XZ Voorburg, The Netherlands b Faculty of Economics and Business Administration, and Institute for Environmental Studies, Free University, De Boelelaan 1105, 1081 HV Amsterdam, The Netherlands ARTICLE INFO Article history: Received 5 May 2005 Received in revised form 8 November 2005 Accepted 9 November 2005 Available online 3 February 2006 Keywords: Environmental accounts Input output modeling Mass balance Material flows Physical accounting System of Environmental and Economic Accounting ABSTRACT A physical input output table (PIOT) provides a framework in which all the physical flows associated with an economy can be recorded. This makes it a valuable tool for environmental economic modeling and accounting. During the 1990's PIOTs were constructed for a number of countries. Subsequently, the PIOT and related physical supply and use tables have been taken up in the System of Environmental and Economic Accounting [UN (United Nations), 2003. System of Environmental and Economic Accounting. United Nations, New York]. This paper reviews PIOTs for the Netherlands, Germany, Denmark, Italy, Finland and the European Union. These studies applied the basic and extended PIOT frameworks. This paper elaborates these frameworks with packaging, residuals (wastes and emissions), recycling and stock changes, in order to create a full PIOT. The production process of the extended and full PIOT is split into structural and auxiliary production processes. Modeling applications and ways of deriving environmentally relevant information from the full PIOT are illustrated using a numerical example. 2005 Elsevier B.V. All rights reserved. 1. Introduction In the past decades, increasing attention for environmental issues has stimulated interest in environmental satellite accounts to complement monetary national accounts. A variety of accounts have recently been summarized in the System of Environmental and Economic Accounting (UN, 2003). The physical input output table (PIOT) is one of these physical accounts. It records all physical flows associated with the economic activities as defined in the System of National Accounts (SNA) (UN, 1968; UN, 1993). The PIOT registers flows of physical products, extraction of materials from nature, the supply and use of wastes and residuals, emissions to nature and stock changes. Furthermore, a PIOT uses the same classification schemes for production activities as the monetary IO table (MIOT). As a result, data from the PIOT and the MIOT can be combined to support environmental economic analysis and policy modeling at the national level. Similar to the MIOT, a PIOT thus serves accounting and modeling purposes. This paper will argue that the most important function of the PIOT is as an accounting framework for environmentally This research was done in the context of the research program Materials Use and Spatial Scales in Industrial Metabolism (MUSSIM), which was funded by the Netherlands Organisation for Scientific Research (NWO). Corresponding author. E-mail addr esses: rhka@cbs.nl (R. Hoekstra), jbergh@f eweb.vu.nl (J.C.J.M. van den Bergh). 0921-8009/$ - see front matter 2005 Elsevier B.V. All rights reserved. doi:10.1016/ j.ecolecon.2005. 11.005

376 ECOLOGICAL ECONOMICS 59 (2006) 375 393 relevant physical flows. As such it is a valuable tool for environmental economic modeling. However, the PIOT also has two advantages in terms of National Accounts which are briefly discussed hereafter. First, similar to the MIOT, the PIOT 1 serves as an integration framework for different data sources. Indeed, data for PIOTs are obtained from a variety of sources, such as energy accounts, waste accounts, production statistics, recycling data, emissions statistics and international trade statistics. The application of the mass balance principle means that a coherent picture emerges from the integration of data sources. This process ultimately results in the homogenization of classification schemes and data collection methods. Furthermore, it allows the identification of errors, such as over- and underestimation, in statistics. A second advantage of the PIOT is that it can improve the monetary statistics of the National Accounts, because it highlights a different (physical) dimension of the same economy. The physical accounts can therefore validate or even improve the existing MIOT. This is particularly the case where the source data for the MIOTs are of a poor quality. 2 For example, the use of intermediate inputs or additions to stocks is often not supported by good source statistics. Monetary estimates can be improved by multiplying mass values from the PIOT by a price per kilogram. Furthermore, physical units make it easier to communicate with process engineers about implicit technological relationships. This paper reviews and extends the PIOT framework. This involves consideration of PIOTs for the Netherlands, Germany, Denmark, Italy, Finland and the European Union. A distinction is made between a basic and extended PIOT framework. The latter splits the production process into a structural production process and an auxiliary production process, as originally proposed by Konijn et al. (1995). A new extension of this framework is proposed, which includes packaging, waste, recycling and stock changes. The resulting full PIOT framework is illustrated with a numerical example. In addition, the relevance of the new framework is shown by illustrating the type of environmentally relevant information and modeling that can be based on PIOTs. The organization of the remainder of this paper is as follows. Section 2 reviews the PIOTs that have been constructed for five European countries and the European Union. In Section 3, the basic PIOT framework is introduced. Section 4 discusses the extended PIOT. In Section 5, this extended PIOT framework is elaborated in order to create a full PIOT. To illustrate the extended PIOT system, a numerical example is presented in Section 6. Section 7 discusses various types of environmental information and indicators that can be derived from an extended PIOT. In Section 8 the use of extended PIOT data in IO modeling is discussed. Section 9 concludes. 1 The integrating function is best performed by the supply and use tables (SUT), but we will not elaborate on this issue here. SUT are related to the input output tables and share similar accounting identities. 2 For some accounts the physical data sources are better than the monetary dimension. For example, the Dutch National Accounts use physical information on energy flows as the basis for their monetary data in the supply and use table. 2. PIOT studies in five countries and the EU Frameworks for the accounting of economic environmental relationships date back to the late 60s and early 70s (Cumberland, 1966; Daly, 1968; Ayres and Kneese, 1969; Isard, 1972; Victor, 1972; Leontief, 1970; Leontief and Ford, 1972). These frameworks have resulted in a range of physical and environmental national accounts, some of which have been adopted in the System of Environmental and Economic Accounting (UN, 2003). Many of the SEEA accounting frameworks employ physical units: e.g. economy-wide MFA, the national accounting matrix with environmental accounts (NAMEA), physical supply and use tables (PSUT), physical IO tables (PIOTs), waste accounts and resource accounts. This paper will not discuss further the relationship between the physical accounts, but rather will focus on the elaboration and application of the physical input output tables (PIOT). PIOTs exist for five countries: The Netherlands, Germany, Denmark, Italy and Finland. Furthermore, a preliminary PIOT for the European Union is based on information from the German and Danish PIOT, scaled up to EU levels. 3 The characteristics of these six projects are summarized in Table 1. Four studies record the flows for 1990 and three for 1995. The Dutch study introduced a PIOT, which we will refer to as an extended PIOT, while the rest used a basic PIOT structure, which is similar to the traditional MIOT structure. The basic and extended PIOTs will be discussed in Sections 3 and 4. Most of the studies in Table 1 are focused solely on the accounting function of the PIOT, while the Dutch and EU studies provide modeling applications as well. 4 The disaggregation of the production structure ranges from 5 to 60 industries. The German, Italian, Finnish and EU PIOTs present total mass of materials and products, while the Dutch and Danish studies also report information for specific materials. For example, the Danish PIOT distinguishes nine material types, including packaging. The primary function of a PIOT is to serve as an economic environmental accounting framework, which can be used in environmental analysis. To fulfill this function three conditions must be met: 1. The PIOT must adhere to economic (IO) modeling assumptions. 2. The PIOT must adhere to National Accounts definitions. 3. The physical flows should be relevant to environmental problems. 3 Tentative efforts to produce an Austrian PIOT are currently taking place at the Department for Social Ecology, Institute of Interdisciplinary Studies of Austrian Universities (IFF), Vienna, Austria. A preliminary PIOT for Austria was constructed for 1983 (Kratterl and Kratena, 1990). Hoekstra (2005) constructs a hybrid IO table for iron and steel and plastics for the Netherlands. A PIOT project is also currently underway in Japan at the National Institute for Environmental Studies (NIES). 4 Giljum and Hubacek (2001) review a number of PIOT models.

ECOLOGICAL ECONOMICS 59 (2006) 375 393 377 Table 1 Summary of characteristics of PIOT studies Characteristics The Netherlands Germany Denmark Italy Finland EU Year for which table is made 1990 1990 and 1995 1990, Air emissions for 1990 92 1995 1995 1990 PIOT type Extended PIOT Basic PIOT Basic PIOT Basic PIOT Basic PIOT Basic PIOT Aim Accounting, Accounting Accounting Accounting Accounting Modeling modeling Production unit (number) Secondary materials Industries (58 and 60 in 1990 and Industries (27) Industries (5) Industries (30) Industries (7) (Differs per material) 1995 respectively) Materials specified Comments Cement, concrete and concrete products; plastics; non-ferro metals; paper and paper products; iron, steel and zinc; energy. Introduces the extended PIOT. Monetary and physical accounts use different classifications. Models of materials use included. Reference Konijn et al. (1995) and Konijn et al. (1997) Total mass; energy; water; and other materials. The study for 1990 distinguishes 3 material groups. This paper inspired subsequent studies. Total mass; animal and vegetable products; stone, gravel and building materials; energy (ton and PJ); metals, machinery, apparatus and means of transport; chemical products and fertilizers; plastics and plastic products; wood, paper and commodities thereof; other commodities; packaging; nitrogen content. Distinguishes many materials and products flows. Stahmer et al. Gravgård-Pedersen (1997) and (1999) Statistische Bundesambt (2001) Total mass; carbon. Aggregate study in terms of industry and material classification. Text is in Italian. Nebbia (2000) Total mass. Work in progress. Focused on: combustion, biological material flows and water content changes. Mäenpää and Muukkonen (2001), Mäenpää (2002) Total mass. A preliminary estimate of a PIOT for the EU, based on the German and Danish PIOTs. Includes PIOT modeling exercises. Giljum and Hubacek (2001) and Hubacek and Giljum (2003) Condition 1 is a general issue concerning IO modeling, whether physical or monetary. IO tables are produced in industry-by-industry, commodity-by-commodity or other dimensions. Konijn (1994) shows that an activity-by-activity IO table, which uses production process as the basis for classification, adheres best to IO modeling assumptions. The theoretical structure that we have chosen in this paper is consistent with this activity-by-activity IO table. Condition 2 is important because the PIOT will be more useful for policy and modeling purposes if it can be compared to the MIOT. The disaggregation is also important: PIOTs that specify below 10 economic sectors will therefore be less useful than the German PIOT, which distinguishes 60 sectors. In our view the Third condition is not sufficiently satisfied by the present generation of PIOTs. Unfortunately, the PIOT studies tend to publish in very aggregated categories which means that the data are not very useful in environmental analysis. For instance, the Dutch study provides detailed information about physical products, but it distinguishes only a few emissions and extraction categories, including the balance of additions and subtractions. Sometimes the problem is not caused by a lack of data, but the authors make unfortunate choices in the aggregation of the publication data. For example, the published German PIOT for 1990 distinguishes only two raw materials: produced and non-produced natural assets. Emissions are recorded solely for landfills, produced and non-produced natural assets. This information is only of limited use for environmental analysis. However the German supply and use tables distinguish nine raw materials and eleven emission categories. 5 This data could have been published in the PIOT but instead were aggregated. The German study for 1995 is a great improvement in this respect: it distinguishes a far greater range of emissions and raw material categories. Waste and extraction categories are fairly aggregated in the Italian and Finnish PIOTs. The Danish study provides one extraction, one recycling and seven emission categories, but more environmental information can be derived from the other PIOTs for specific materials. The focus on aggregate material flow data is also common in the related field of economy-wide material flow analysis (MFA) (Adriaanse et al., 1997; WRI, 2000). These publications suggest that MFA indicators, which 5 A separate physical account for emissions to air is also presented in the German study.

378 ECOLOGICAL ECONOMICS 59 (2006) 375 393 aggregate different bulk materials, irrespective of their environmental impact, can be used as measures of sustainability. Perhaps this is the reason why some PIOTs focus (too) much on accurately estimating all physical flows in a material balance, instead of prioritizing those that are environmentally relevant. 6 For the PIOT to become a useful part of the environmental economic toolkit, it should focus on its function as a data source for environmental analysis. This will only be achieved if it is seen to contain sufficiently disaggregated data on environmentally relevant physical flows. 3. The basic PIOT 7 The starting point for our theoretical framework is the MIOT and basic PIOT shown in Table 2. The economy is split into production processes. We assume that both tables are consistent with the homogeneous production assumptions of the IO model: each production process creates one commodity and each commodity is produced by one production process. Moreover, in addition to the principal product, some of the production processes generate wastes. The residual by-products are recorded as negative inputs, which is consistent with the by-product IO model proposed by Stone (1961). If the by-product method is not used, the IO models show that increases in demand for the by-product would lead to increased demand for the principal product. For example, a producer of metal products would increase output of these products if demand for scrap increased. The basic PIOT accounts for all materials, products and waste flows associated with economic processes. Table 2 distinguishes α production processes, η types of primary inputs, β types of raw materials and γ types of residuals. 8 The basic PIOT and MIOT tables specify the same production processes and final demand categories. The intermediate matrix Z and Z and the final demand vector Y and Y have the same dimensions (bold font is used for physical variables while normal font is used for monetary variables). However, the MIOT contains output data for all production processes, while the PIOT only records the flows of goods that have a physical dimension. 9 However, 6 An example is the Finnish PIOT (Mäenpää and Muukkonen, 2001; Mäenpää, 2002). It distinguishes no less than 21 different foodstuffs to evaluate the human metabolic balance. Although this is a thorough piece of work, we would argue that there are far more urgent problems that should be tackled with the PIOT framework. 7 The basic PIOT presented in this section differs slightly from those of the studies reviewed in Table 1, in that the PIOT presented here records the supply of residuals as negative inputs. 8 The term residual is adopted because it is used in the SEEA. It is used synonymously with the terms emissions and wastes, which indicate any solid, liquid or gas waste by-product of production processes. These may be recycled, landfilled, incinerated or emitted to nature. 9 In the rest of the paper it is assumed that services do not produce any physical products. Table 2 A MIOT and basic PIOT Production processes 1 α Final demand Output MIOT Production Processes 1 α Z Y q Primary Inputs 1 η K Use of Residuals 1 γ R Supply of Residuals 1 γ W Total q Basic PIOT Production processes 1 α Z Y q Raw materials 1 β D Use of residuals 1 γ R Supply of residuals 1 γ W Emission to nature 1 γ E Total q Monetary balance ZVd i a þ KVd i D þ RVd i g W Vd i g ¼ Zd i a þ Y ¼ q ð1þ Mass balance Z Vd i a þ DVd i h þ RVd i g W Vd i g þ EVd i g ¼ Zd i a þ Y ¼ q ð2þ Here i α, i β, i γ and i η are the appropriate summation vectors. the PIOT records all residual by-products that are supplied to other production processes in the economy (W) or emitted to nature (E). The MIOT records only the supply of residuals that have a price (W). The use of residuals by production processes is recorded in matrix R and R. The flows of wastes to the recycling, landfilling and incineration sectors as well as emissions to nature are further specified in the extended PIOT of Section 5. Finally, both tables have different non-industrial inputs. For the MIOT these are the primary inputs K, such as labor and capital depreciation, while for the basic PIOT it is the extraction of raw materials that is recorded in matrix D. 10 Note that the physical flows recorded in matrices D and E are not all environmentally relevant. Many inputs and outputs have no or negligible environmental impact. For example, the use of oxygen in the combustion process of fossil fuels is not considered to be an environmental problem. Similarly, the emission of water in this combustion process is also not a significant issue. Nevertheless, these flows are part of PIOT that provides a full description of all material flows. The MIOT and basic PIOT exhibit monetary and mass balance, respectively (see Eqs. (1) and (2) below the MIOT and basic PIOT). The relationship between the basic PIOT and MIOT is expressed in Eq. (3). This is a straightforward 10 The SEEA distinguishes two types of raw materials: natural resources and ecosystem inputs. Natural resources cover mineral and energy resources, and biological resources. Ecosystem inputs cover air and gases necessary for combustion and the water to sustain life. (UN, 2003, 2 5, 2 6). The distinction could easily be introduced but seems fairly arbitrary, which is why it is not adopted in this article.

ECOLOGICAL ECONOMICS 59 (2006) 375 393 379 Production processes Final demand Production processes Intermediate use Final demand Monetary accounts Secondary materials Production processes Final demand Secondary materials Use for transformation Final use Final use Physical accounts Primary materials Use for transformation Final use Final use Fig. 1 The MIOT-extended PIOT framework proposed by Konijn et al. (1995). relationship: value equals mass multiplied by price per unit mass. 11 ZV¼ Z Vd p â Y V¼ Y Vd p â W V¼ W Vd p ĝ RV¼ RVd p ĝ Vector p α includes the prices of goods per unit mass, while p γ is the vector of waste prices per unit mass. 12 Note that there is no relationship between the MIOT and basic PIOT for production processes that do not produce a physical commodity. Therefore, the equations for Z and Y only hold for physical commodities. The monetary values of the services should be obtained directly from the MIOT. 4. The extended PIOT This section discusses the extended PIOT framework proposed by Konijn et al. (1995). The monetary and physical accounts are shown in Fig. 1. For the monetary accounts an activity-byactivity MIOT is used. The physical accounts, which will be referred to as the extended PIOT, are split into several parts. The main innovation in Fig. 1 is the distinction between use for transformation and final use of materials. The former refers to physical inputs that are converted into other physical products, e.g., refining of crude oil to petroleum would entail ð3þ use for transformation. Final use is similar to the final demand concept in the monetary accounts, e.g., burning of petroleum by the transport sector would be deemed final use. This is the final market of a product, where it is consumed. 13 In the empirical applications presented in Konijn et al. (1995), the monetary and physical economies use different bases, which is also why there is no link between the monetary intermediate inputs and the physical intermediate flows. The MIOT has an activity-by-activity structure, while the intermediate final use has an activity-by-secondary materials classification, and the use for transformation has a secondary materials-by-secondary materials structure. In the applications provided, the disaggregation of the secondary materials is greater than that of the activity classification. This has the advantage of allowing the construction of detailed physical accounts, without having to adapt the MIOT. However, from a theoretical perspective, there is no reason to use different classification schemes for the monetary and physical economies. The extended PIOT is particularly useful in the analysis of physical products. However, other physical flows which are relevant for environmental analysis, such as residuals, stock changes and packaging are not included in the theoretical framework. Recycling is included in the empirical applications presented by Konijn et al. (1995) in so far as they are traded goods in the economy. 11 Note that mass is not necessarily the same as volume in the National Accounting sense. We are grateful to a reviewer for pointing out that the disaggregation at which this relationship holds should be far more detailed than the sector disaggregation of the PIOTs published to date. See Weisz and Duchin (in press) for further discussion of the influence of heterogeneous sectors. 12 In this paper the prices are assumed to be the same for all types of consumers, although they are likely to vary in a real world situation. 13 The final destination of value or mass is different. For example, goods and services are supplied to consumers. (Final) consumption of the goods and services leads to loss of value. In the MIOT, this is therefore the final destination of value. However, physical consumption does not exist. Physical goods are used for the services that they provide. However, when they lose their value, they do not lose their mass. The final destination of physical flows is, therefore, the disposal of product waste (Ayres et al., 2005). In the case of recycled flows there is no source and destination for the accounting period.

Table 3 The full PIOT framework Input/Output Structural production process Auxiliary production process Total Production processes Raw materials Use of residuals Supply of residuals Emissions to nature Stock changes 1 α 1 α Packaging Recycling Other goods Landfilling Incineration Other services Packaging Recycling Other goods Recycling Other goods Z str Z aux Incineration Landfilling α Other services 1 Materials Materials D str D aux β Materials Recycling R str R aux Other goods γ Materials Recycling W str W aux Other goods γ Materials Recycling E str E aux Other goods γ Materials Recycling S str S aux Other goods γ Materials Total q 0 Landfilling Incineration Other services 380 ECOLOGICAL ECONOMICS 59 (2006) 375 393 Input/Output Investment goods Consumption Total 1 α 1 2 Packaging Recycling Other goods Landfilling Incineration Other services Non-durable goods Durable goods

Table 3 (continued) Input/Output Investment goods Consumption Total Production processes Raw materials Use of residuals Supply of residuals Emissions to nature Stock changes 1 α 1 2 Packaging Recycling Other goods Landfilling Incineration Other services Non-durable goods Durable goods Recycling Other goods F C q Incineration Landfilling α Other services 1 Materials Materials D cons d β Materials Recycling R cons r Other goods γ Materials Recycling W inv W cons w Other goods γ Materials Recycling E cons e Other goods γ Materials Recycling S inv S cons s Other goods γ Materials Total 0 0 0 ECOLOGICAL ECONOMICS 59 (2006) 375 393 381

382 ECOLOGICAL ECONOMICS 59 (2006) 375 393 5. The full PIOT In this section, the extended PIOT framework, as initially proposed by Konijn et al. (1995), is elaborated. While the extended PIOT is designed to describe the physical flows of products in the economy, we will present a new framework which includes a full description of the physical flows including recycling, stock changes, packaging, wastes and emissions. This elaborated framework will be referred to as the full PIOT. Different terminology is introduced. Use for transformation is replaced by the structural production process, while intermediate final use is called the auxiliary production process. These terms are chosen because they reflect the technological functions of materials in the production processes. The terms secondary materials and primary materials are replaced by intermediate inputs and raw materials, respectively. This is to avoid confusion with the more common use of the term secondary materials, which is often used to refer to recycled materials. Table 3 shows a full PIOT framework, with α production units, β types of raw materials, γ types of residuals, and δ types of stock changes. Three categories of goods (packaging, recycling and other goods) and three categories of services (landfilling, incineration and other services) are distinguished. Consumption is split into durable and non-durable goods. The relationship between the basic and full PIOT is provided by Eq. (4). Variables of the basic PIOT are given on the left-hand-side; full PIOT variables on the right-hand-side. The structural and auxiliary inputs are denoted by subscripts str and aux, respectively. The final demand vector (Y) of the basic PIOT is the aggregate of the investments (F) and consumption (C) matrices. Z ¼ Z str þ Z aux Y ¼ Fd i a þ Zd i a D ¼ D str þ D aux R ¼ R str þ R aux W ¼ W str þ W aux E ¼ E str þ E aux Four types of physical inputs and outputs of a production process can be distinguished: the structural production process; auxiliary production process; investment goods; and unwanted physical inputs and outputs. The structural production process encompasses all material flows that are associated with making the physical good. The structural production process inputs are: packaged product inputs (Z str ); raw materials (D str ) and use of residuals (R str ). The structural outputs are the product of the production process, which is sold to other production processes (Z str, Z aux and F) or to consumers (C). However, not all structural inputs are converted to products. Residuals may result from the structural production process, which can be supplied to other production processes (W str ) and emitted to nature (E str ). Stock changes are also possible (S str ). The structural production process is best illustrated by an example. In the production of cars, the structural inputs would be metal plating and engine parts, while the structural outputs would be the car itself as well as any scrap metal from the body parts. ð4þ The mass balance of the structural production process equates the column sum to its row sum, as is shown in Eq. (5). Note the special attention for packaging in the structural production process. Packaging is viewed as a structural input in the production process because it accompanies the finished product to the consumer. Since services do not produce physical commodities, there are no structural inputs for these production processes. The mass balance equation for the structural production process is: Z str V d i a þ D str V d i h þ R str V d i g W str V d i g E str V d i g S str V d i y ¼ Z str d i a þ Z aux V d i a þ Fd i a þ Cd i a ¼ q The auxiliary production process includes all physical flows that are required to facilitate the production process, except those that are used in the structural production process. The physical inputs are not incorporated in the physical structure of the end product but they are used for other reasons. They also differ from investment goods because they are used within the accounting period of the IO table. The auxiliary production process uses packaged products (Z aux ) and raw materials (D aux ). Residuals (R aux ) can be used by, for example, the landfilling and incineration sectors. These waste conversion processes store or transform these residual flows. The auxiliary outputs are residuals (W aux and E aux ) and stock changes (S aux ). An example of an auxiliary production process could be a combustion process which requires energy and oxygen (auxiliary) inputs, while the CO 2 is considered an auxiliary output. The mass balance equation for the auxiliary production process is: Z aux V d i a þ D aux V d i h þ R aux V d i g W aux V d i g E aux V d i g S aux V d i y ¼ 0 Investment goods are required in production processes. They facilitate production, but do not end up in the products, which makes them similar to auxiliary inputs. However, investment goods are used over multiple periods. The physical balance for investment goods is shown in Eq. (7). Packaged investment goods are supplied by other production processes (F). However, in the matrix F, the rows for recycling and packaging are negligible. Similarly, raw materials or residuals are not used in the investment goods account. 14 On the output side, capital goods are written off and supplied to others in the economy (W inv ). These capital goods may be sold to other economic sectors, or to the recycling, landfilling or incineration production processes. Stock changes (S inv ) are also included in the investment account: F V inv d i a W inv V d i g S inv V d i y ¼ 0 Unwanted materials, which have no technological function in the production process, can enter or leave a production process. The best example of an input that has no technological function is the packaging layer that accompanies all structural, auxiliary and investment inputs. This layer is mostly needed in the transportation and storage phase, but has no role to play in the production process itself. 14 Some of the raw material inputs associated with capital goods are recorded as auxiliary inputs. For example, if machines use fossil fuels for energy, a certain amount of oxygen is required. This is registered as auxiliary inputs. ð5þ ð6þ ð7þ

ECOLOGICAL ECONOMICS 59 (2006) 375 393 383 Final consumption also adheres to mass balance, as is shown in Eq. (8). Consumers buy packaged products from production processes (C). Some of these products, such as fossil fuel-driven cars, require raw materials such as oxygen (D cons ). Small amounts of the residuals may also be used (R cons ) by consumers. The outputs of consumption are residuals that are supplied to other economic categories (W cons ) or emitted to the environment (E cons ), and stock changes (S cons ). For the non-durable goods the stock changes will be minimal. C Vd i a þ D cons V d i h þ R cons V d i g W cons V d i g E cons V d i g S cons V d i y ¼ 0 ð8þ The mass balance for the entire economy is defined in Eq. (9). The raw material input of the economy (d i β ) is matched to the output: emissions (e i γ ) and stock change (s i δ ): dvd i h ¼ evd i g þ svd i y ð9þ where d ¼ D str d i h þ D aux d i h þ D cons d i h r ¼ R str d i g þ R aux d i g þ R cons d i g w ¼ W str d i g þ W aux d i g þ W inv d i g þ W cons d i g e ¼ E str d i g þ E aux d i g þ E cons d i g s ¼ S str d i y þ S aux d i y þ S inv d i y þ S cons d i y The waste processing flows of Table 3 require further clarification. Residuals are generated by production processes and consumers. One portion of these wastes is emitted to the environment (E str, E aux, E inv and E cons ), while the other part is supplied to the economic sectors (W str, W aux, W inv and W cons ). The latter residual flows are used by production processes as structural inputs (R str ) or auxiliary inputs (R aux ), while a small portion may be used by consumers (R cons ). The supply of residuals by the economy is therefore equal to the use of residuals, as shown in Eq. (10): w ¼ r ð10þ The residuals that are used as structural inputs (R str ) are recycling flows. They are used to be embodied in physical goods. The recycling of residuals can occur in one of two ways. They can be used directly in production processes, e.g. scrap metal used to produce primary metal. Optionally, they can be supplied to a recycling sector that treats the residual flow so that it may be reused, e.g. cars may be supplied to shredder companies. The auxiliary input of residuals (R aux ) is the use of residuals from the auxiliary production process. These flows include the input of residuals in the landfilling and incineration production processes. 15 In the case of landfilling, the output is a change in the stock of materials. For the incineration sector, the residuals are converted into emissions to nature and perhaps some residual ash. The latter would then serve as an input for the landfilling sector. 16 15 Landfilling and incineration are regarded as services. The recycling sector produces a good output because it makes a physical product that is supplied to the market. 16 The description of waste and recycling flows could be elaborated even further than done in Table 3, as the use of residuals rows do not specify the origin of the residuals that are used. Similarly, the supply of residuals does not specify to whom the residuals are supplied. This information can easily be incorporated in the rows of the PIOT or in separate satellite accounts for recycling. 6. The full PIOT: a numerical example A numerical example of a full PIOT is shown in Table 4. The corresponding MIOT, hybrid IO table and price information are provided in Appendix A. The technological and demand relationships are hypothetical. Ten production processes make physical products (mined metal, mined fossil fuel, basic metal, basic plastics, metal packaging, plastic packaging, recycled metal, recycled plastic, machines and objects). Three types of services are provided: landfilling, incineration and business services. Each of the 10 structural production processes of physical goods is discussed briefly (abbreviations are provided in brackets): 1. Mined metal (MM) is produced by extracting unmined ore. The structural input is unmined ore. Structural outputs are mined metal and mining debris. 2. Mined fossil fuel (MFF) is produced by extracting unmined fossil fuel. The mined fossil fuel is packaged in metal barrels. Structural inputs are unmined fossil fuel and metal packaging. Structural outputs are mined fossil fuel and mining debris. 3. Basic metal (BM) is produced by using mined metal and recycled metal. Structural inputs are mined metal and recycled metal. Structural outputs are basic metal and waste metal. 4. Basic plastic (BP) is produced by using mined fossil fuel and recycled plastic. It is packaged in metal. Structural inputs are mined fossil fuels, recycled plastics and plastic packaging. Structural outputs are basic plastic, waste plastic and waste metal packaging (from the mined fossil fuel deliveries). 5. Metal packaging (MP) is produced using basic metal. The structural input is basic metal. The structural output is metal packaging. 6. Plastic packaging (PP) is produced using basic plastic. The structural input is basic plastic. The structural output is plastic packaging. 7. Machines (Mch) are produced using basic metal and plastics. They are packaged in metal and plastic. Structural inputs are basic metal and basic plastics. Structural outputs are machines, waste metal, waste plastic and waste metal packaging (from the basic plastic inputs). 8. Objects (Obj) are produced using basic metal and plastics. They are packaged in metal and plastic. Structural inputs are basic metal and basic plastics. Structural outputs are objects; waste metal; waste plastic; and waste metal packaging (from the basic plastic inputs). 9. Recycled metal (RM) is produced using waste metal and waste metal packaging. Structural inputs are waste metal and waste metal packaging. The structural output is recycled metal. 10. Recycled plastic (RP) is produced using waste plastic. The structural input is waste plastic. The structural output is recycled metal. The auxiliary production process is similar for most production processes: fossil fuels are used for energy. This combustion process requires oxygen and emits water and carbon dioxide.

Table 4 A numerical example of a full PIOT Structural production process MM MFF BM BP MP PP Mch Obj RM RP LF IN BS Mined metal (MM) 1843 Mined fossil fuel (MFF) 1073 Basic metal (BM) 704 900 700 Basic plastics (BP) 220 550 660 Metal packaging (MP) 474 130 50 50 Plastic packaging (PP) 100 100 Machines (Mch) Objects (Obj) Recycled metal (RM) 691 Recycled plastic (RP) 390 Landfilling (LF) Incineration (IN) Business services (BS) Raw materials Unmined ore 3686 Unmined fossil fuel 11,850 Oxygen Use of residuals Waste metal 410 Waste plastic 390 Waste metal packaging 281 Waste plastic packaging Waste machines Waste objects Supply of residuals Waste metal 230 100 80 Waste plastic 65 300 220 Waste metal packaging 98 20 50 60 Waste plastic packaging Waste machines Waste objects Emissions to Nature Mining debris 1843 7110 Carbon dioxide Water Stock Changes Waste metal Waste plastic Waste metal packaging Waste plastic packaging Waste machines/ machines Waste objects/objects Total 1843 5214 2304 1430 704 200 1150 1150 691 390 0 0 0 384 ECOLOGICAL ECONOMICS 59 (2006) 375 393

Table 4 (continued) Auxiliary production process MM MFF BM BP MP PP Mch Obj RM RP LF IN BS Mined metal (MM) Mined fossil fuel (MFF) 406 1043 507 286 39 11 220 220 38 21 251 Basic metal (BM) Basic plastics (BP) Metal packaging (MP) Plastic packaging (PP) Machines (Mch) Objects (Obj) Recycled metal (RM) Recycled plastic (RP) Landfilling (LF) Incineration (IN) Business services (BS) Raw materials Unmined ore Unmined fossil fuel Oxygen 1261 3244 1577 890 120 34 684 684 118 67 1352 781 Use of residuals Waste metal Waste plastic 195 Waste metal packaging 423 Waste plastic packaging 200 Waste machines 727 Waste objects 739 Supply of residuals Waste metal Waste plastic Waste metal packaging 37 95 46 26 4 1 20 20 3 2 23 Waste plastic packaging Waste machines Waste objects Emissions to Nature Mining debris Carbon dioxide 1157 2974 1446 816 110 31 628 628 108 61 1239 716 Water 473 1218 592 334 45 13 257 257 44 25 507 293 Stock Changes Waste metal Waste plastic Waste metal packaging 423 Waste plastic packaging Waste machines/ 727 machines Waste objects/objects 739 Total ECOLOGICAL ECONOMICS 59 (2006) 375 393 Investment goods Consumption (continued on next Output page) MM MFF BM BP MP PP Mch Obj RM RP LF IN BS Non-durable Durable Mined metal (MM) 1843 385

Table 4 (continued) Investment goods Consumption Output MM MFF BM BP MP PP Mch Obj RM RP LF IN BS Non-durable Durable Mined fossil fuel (MFF) 1100 5214 Basic metal (BM) 2304 Basic plastics (BP) 1430 Metal packaging (MP) 704 Plastic packaging (PP) 200 Machines (Mch) 74 191 93 52 3 1 40 40 3 2 2 2 230 417 1150 Objects (Obj) 37 95 46 26 1 0 20 20 1 1 2 2 115 781 1150 Recycled metal (RM) 691 Recycled plastic (RP) 390 Landfilling (LF) Incineration (IN) Business services (BS) Raw materials Unmined ore 3686 Unmined fossil fuel 11,850 Oxygen 3422 14,235 Use of residuals Waste metal 410 Waste plastic 585 Waste metal packaging 704 Waste plastic packaging 200 Waste machines 727 Waste objects 739 Supply of residuals Waste metal 410 Waste plastic 585 Waste metal packaging 5 12 6 3 0 0 3 3 0 0 0 0 15 100 52 704 Waste plastic packaging 10 25 12 7 0 0 5 5 0 0 0 0 30 104 200 Waste machines 46 119 58 33 2 1 25 25 2 1 2 2 143 272 727 Waste objects 23 59 29 16 1 0 13 13 1 0 2 2 71 510 739 Emissions to Nature Mining debris 8953 Carbon dioxide 3138 13,053 Water 1284 5343 Stock Changes Waste metal 0 Waste plastic 0 Waste metal packaging 423 Waste plastic packaging 0 Waste machines/ 18 47 23 13 1 0 10 10 1 0 1 1 57 91 1000 machines Waste objects/objects 9 24 12 7 0 0 5 5 0 0 1 1 29 170 1000 Total 386 ECOLOGICAL ECONOMICS 59 (2006) 375 393

ECOLOGICAL ECONOMICS 59 (2006) 375 393 387 The landfilling and incineration sectors are different. Waste metal packaging, waste machines and objects are landfilled in storage sites, as indicated by the changes in the auxiliary stocks shown in Table 4. 17 For the incineration sector, waste plastics and waste plastic packaging are burned in the presence of oxygen to produce water and carbon dioxide. Machines and objects are used as investment goods and consumer products. A distinction is made between non-durable consumption, such as fossil fuels, while machines and objects are assumed to be durable commodities. Table 4 will be used in the following sections. 7. The full PIOT: deriving environmental information The full PIOT provides a wide variety of opportunities to extract environmental information. A number of examples, based on the information in Table 4 and Appendix A are provided in this section. Full PIOTs can provide information about various types of environmental pressure, composition of products, element cycles in the economy, dematerialization indicators, and international indicators. 7.1. Environmental pressure indicators The environmental inputs required by the economy are represented by the resource use vector d in Table 3. In Table 4, the extraction of metal and fossil fuel are environmental problems because of the depletion of these resources. Emissions to nature are recorded in vector e in Table 3. The numerical example of Table 4 includes carbon dioxide (CO 2 ) emissions, which lead to global warming. Mining debris can cause local environmental problems. The stock changes in Table 4, vector s in Table 3, are also of interest to environmental policy. For example, the increase in the wastes that are stored in landfills is recorded as changes in stocks. Furthermore, the other stocks in the economy will at some stage in the future, end up in landfills or be emitted to nature. Table 5 provides a number of examples of physical flows that are of interest to environmental policy. Note that the detailed structure of the full PIOT makes it possible to specify these indicators separately for each structural production process, auxiliary production process, investment goods account or consumers. These could serve as indicators for targeted policies. 7.2. Composition of products 17 Negative values indicate increases in stocks. 18 Note that this calculation is always valid on the element-level but not necessarily on the chemical/compound level, because chemicals or compounds can change in production processes. Nevertheless for many production processes a compound or compound balance can be stated. Table 5 Environmental indicators Environmental problem Assuming that the composition of the raw materials, and residuals are known, the composition of the product can be derived. Multiplying the mass of the inputs and outputs of the structural production process by the composition yields a remainder, which is embodied in the product. 18 The compositions of materials and the derived composition of products are shown in Appendix A (Tables A.4 and A.5, respectively). The composition (Iron-Fe, Carbon-C, and Hydrogen-H) of products and their packaging layer is provided in Fig. 2. This information indicator of product composition illustrates that a PIOT can provide a link between macroeconomic material flows and microeconomic concept such as products. The product composition found by the PIOT can be compared with studies that have estimated the composition of products through direct means (Hansen, 1995). Note that in reality relatively few of the elements of the periodic table are used in significant quantities in physical goods. The composition of the raw materials and emissions could also be added to the figure, as well as an analysis of the direct and indirect input requirements of these substances. Note that this is a very ambitious application of the full PIOT. It has never been attempted before but would provide an invaluable link between materials, emissions and products. Constructing such a PIOT would allow the thorough analysis of substitution and product composition changes. 7.3. Element cycles in the economy Environmental scientists construct element cycles, such as the carbon or nitrogen cycle, which provide insights into the sources, sinks and flows of certain elements. 19 Element cycles can also be produced for the economic system. Fig. 3 depicts the economic carbon cycle for the numerical example. The full PIOT material and product flows of Table 4 are multiplied by the chemical composition data for carbon from Appendix A. However, in the case of mined fossil fuel the composition should be corrected for a 10% metal packaging layer. In the case of machines and objects there a 5% metal packaging layer and a 10% plastic packaging layer. A PIOT for nitrogen already exists for Denmark (Gravgård-Pedersen, 1999). 7.4. Dematerialization indicators kilotons Metal ore depletion 1843 Fossil fuel resource depletion 4740 CO 2 emissions 13,053 Landfilled waste 1889 Incinerated waste 395 Mining debris 8953 Environmental flows can be related to monetary indicators. For example, the environmental problems that were noted in Table 5 can be evaluated per unit of value added, for each individual production process. Fig. 4 illustrates this by calculation of the CO 2 per unit of value added (labor payments plus capital costs), which are taken from Table A.2, in Appendix A). In the numerical example, the mined metal, mined fossil fuel, basic metal, basic plastic and 19 The mystery of the missing CO 2 sink, which was an issue in the initial stages of the global warming research, is an example where constructing these cycles was important.

388 ECOLOGICAL ECONOMICS 59 (2006) 375 393 120% 100% Packaging Percentage Fe C H 80% 60% 40% 20% 0% MM MFF BM BP MP PP Mch Obj RM RP Average Physical goods Product Fig. 2 Composition of products and packaging layers. Note: abbreviations of physical goods were introduced in Section 6. incineration exhibit high values for CO 2 emissions per unit of value added. For the latter production process this is not surprising: the core business of the incineration sector is the combustion of carbon-containing plastics. Although the authors are not advocates of aggregated economy-wide MFA indicators such as the Direct Materials Inputs (DMI), Direct Materials Consumption (DMC) and Total Materials Requirements (TMR), it should be noted that these can be derived from a PIOT. 7.5. International indicators In the basic and full PIOT, presented in this paper, a closed economy has been assumed. The frameworks can be easily expanded to include imports and exports. Indicators of the physical trade balances can be constructed using IO models based on full PIOT data with international trade flows (Giljum and Hubacek, 2001). We have not chosen to expand our example with imports and exports because it is fairly straightforward exercise which would complicate the example significantly. 8. The full PIOT: environmental modeling 20 In this section, two IO modeling applications are discussed as applications of the full PIOT: impact analysis and imputation of primary inputs. Note however that there are many other 20 Recently, a number of papers on PIOT modeling have emerged. Hubacek and Giljum (2003) use the EU-15 PIOT to model land appropriation. This paper led to a number of theoretical papers which discuss physical IO analysis further. Suh (2004a,b), Dietzenbacher (2005), Dietzenbacher et al. (2005) and Weisz and Duchin (in press) comment on the modeling techniques proposed in the aforementioned paper. A rebuttal to Suh (2004a) can be found in Giljum et al. (2004). Giljum and Hubacek, 2004 compare several PIOT models to appropriate primary inputs and wastes. On another note, Hoekstra (2005) and Hoekstra and van den Bergh (2004) use structural decomposition analysis (SDA) to construct forecasting and target analysis models. modeling possibilities using IO or other economic modeling tools. The models in this section are referred to as hybrid models. A hybrid model uses different units to measure the different outputs. For example, fossil fuels will be measured in Joules while primary plastics are measured in kilograms. The main advantages of this model are this provides a better description of the technological relationship between input and output and that the model adheres to mass and energy balance (Miller and Blair, 1985; Hoekstra, 2005). Eq. (11) presents the hybrid IO model. q* ¼ðI A*Þ 1 d y* ð11þ The data for this model are provided in Appendix A (Table A.1). The unit of fossil fuels is GJ (GigaJoules), because it is used for energy purposes. However, in the production of basic plastic, the natural unit of fossil fuels is kilotons, because they are used as a structural input. Theoretically, these two functions of fossil fuels should be represented on separate rows of the hybrid table. However, since the GigaJoule/ kilogram ratio is constant for this example, the result is the same. 21 8.1. Impact analysis Impact analysis is a projection method in IO, which is usually used to assess the influence of final demand changes on output. However, technological changes can also be analyzed in PIOT-based impact analyses. Table 6 presents five examples of impact analyses using the numerical example. The percentages presented are compared with those in the base case, which are recorded in Table 6. A number of interesting changes in the economy are considered in the impact analyses. The resulting change in the full PIOT data is calculated assuming the changes in the parameters under 21 For this numerical example, uniform prices are assumed. The monetary and hybrid-unit models will therefore yield equivalent results. An IO model based only on the physical data of the PIOT would yield incorrect results, because the demand for services cannot be measured in kilotons.

ECOLOGICAL ECONOMICS 59 (2006) 375 393 389 Environment Unmined fossil fuel 4059 Economy Basic plastic 223 56 1113 334 17 1058 Recycled plastic 278 Other 1317 Carbon dioxide 17 1317 Incineration 338 167 296 812 Mined fossil fuel 82 Investments 89 372 153 Landfilling 365 61 856 Consumers 212 856 71 Solid waste/stock change Fig. 3 Carbon cycle in the economy. investigation. After this the data is put back in Eq. (11) to obtain the results. The five impact analyses are: 1. Consumption of machines. The final consumption of machines increases by 20%. This demand leads to increases in all environmental indicators. 2. Consumption of services. The final consumption of business services increases by 20%. The environmental pressures do not increase as greatly as in the first impact analysis, while the value added increase is greater. The increases in the quantities of wastes landfilled and incinerated are small. 3. Recycling. The use of recycled metal increases by 50% in the basic metals production process. These inputs replace mined metal in the production process, as is illustrated by the 19% decrease in metal depletion for basic metals. Furthermore, the increased demand for residuals also decreases the quantity of waste that is landfilled by 18.5%. 4. Substitution. The plastic content of machines increases from 20% (see Fig. 2) to 30%. The metal content therefore diminishes, while it is assumed that the weight of the machines remains the same. The results show how structural changes can shift environmental burdens. The depletion of metals, carbon dioxide emission and incineration decreases, but the other environmental indicators increase. 5. Miniaturization. Machines are assumed to become 10% lighter while maintaining their functionality. Again the environmental indicators decrease across the board. Note that these are only a couple of examples of impact analyses and that there are far more possibilities of using full PIOT data in this type of analysis. 8.2. Imputation to final demand Traditional IO models allow for the imputation of labor and capital inputs to final demand. Similarly, raw material requirements and emissions can be imputed to final demand by using the Eq. (12) U ¼ ðd str þ D aux Þd q * 1 Þd ði A*Þ 1 d y*d i a ð12þ W ¼ ðe str þ E aux Þd q * 1 Þd ði A*Þ 1 d y*d i a where Φ is the set of direct and indirect raw material coefficients, and ψ is the matrix for emission coefficients. Table 7 presents the results for the environmental indicators. Table 6 Impact analysis on environmental indicators Metal depletion (%) Fossil fuel depletion (%) CO 2 emissions (%) Additions to landfills (%) Incineration of waste (%) Mining debris (%) Value added (%) Consumption of 3.7 2.4 2.4 3.4 4.6 2.6 2.3 machines Consumption of 0.2 1.1 1.3 0.2 0.0 0.9 5.9 services Recycling 19.0 1.4 1.6 18.5 0.0 5.1 0.3 Substitution 3.5 1.8 0.4 1.8 6.3 0.7 0.0 Miniaturization 4.6 2.4 2.2 4.1 4.4 2.8 0.2 Note: In all these impact analyses, investment goods levels and prices are assumed to remain constant.

390 ECOLOGICAL ECONOMICS 59 (2006) 375 393 CO 2 /value added (ktons/euro) 2.0 1.5 1.0 0.5 0.0 MM FFM BM BP MP PP Mch Obj RM Goods and Services RP LF IN BS Average Fig. 4 CO 2 emissions per unit value added. Note: abbreviations of physical goods were introduced in Section 6. Note that the coefficients shown in Table 7 are ratios of two units. The units of the numerator are kilotons, as indicated by the last column of the table. The denominator is measured in the natural units of the production processes, shown in the last row. For example, for the service unit production process, an increase in final demand of 1 business service unit leads to an increase of 0.04 ktons of mining debris. This is, of course, the effect of indirect demand, because there is no direct mining debris waste created in the production process of business services. 9. Conclusions This paper has discussed the physical input output table (PIOT), which records all physical flows associated with economic activities. PIOT studies for five different countries and the EU were reviewed. In this paper we argue that the primary reason to construct PIOTs is the application in environmental economic analyses. However, they can also lead to improvements in the physical and monetary national accounts and statistics. Three variations of the PIOT were presented in the paper: the basic, extended and full PIOTs. The basic PIOT uses a similar structure as the monetary input output table (MIOT). The extended PIOT splits the production process into two distinct parts: structural and auxiliary production processes, but it focused on the product flows within the economy. In this paper we therefore elaborate the extended PIOT structure to incorporate all physical flows. Residuals, packaging, recycling, landfilling, incineration, stock changes are all included in the full PIOT. In the remainder of the paper the use of the full PIOT as an accounting and modeling tool is illustrated using a numerical, hypothetical example. This example covered ten production processes generating physical products and three types of services. The resulting PIOT was used in the subsequent illustrations of how PIOTs can yield environmental information and indicators. Various environmental indicators can be supported, ranging from metal ore depletion to greenhouse gas emissions. In addition, PIOTs allow the linking of micro and macro level descriptions of material flows, the portrayal of element cycles, the identification of dematerialization mechanisms and indicators, and the study of international physical flows. Finally, the use of full PIOT data in IO modeling applications was illustrated. In particular, impact analysis and imputation to final demand were considered. Table 7 Imputation of environmental indicators to final demand Metal depletion Fossil fuel depletion CO 2 emissions Landfilled waste Incinerated waste Mining debris Natural units MM 1.02 0.50 0.84 0.02 1.42 ktons MFF 0.02 0.01 0.02 GJ BM 0.84 0.94 1.58 0.03 1.59 ktons BP 0.19 2.51 1.90 0.08 0.02 2.20 ktons MP 0.01 0.01 0.02 0.02 m 2 PP 0.03 0.02 0.02 m 2 Mch 0.01 0.03 0.04 0.03 units Obj 0.01 0.03 0.04 0.03 units RM 0.01 0.13 0.22 0.11 ktons RP 0.01 0.13 0.22 0.11 ktons LF ktons landfilled IN 3.14 ktons incinerated BS 0.05 0.08 0.04 business service units units ktons ktons ktons ktons ktons ktons

ECOLOGICAL ECONOMICS 59 (2006) 375 393 391 Appendix A Data used in the numerical example In this appendix, the hybrid table, the monetary table, price information, and the composition of the numerical example are provided. Table A.1 shows the hybrid IO table, for which each row is measured in its own natural unit. The MIOT is shown in Table A.2. The link between the two is provided by the prices provided in Table A.3. The composition of the raw materials and residuals are shown in Table A.4. These values can be used, in combination with the full PIOT, to calculate the composition of the physical commodities shown in Table A.5. Table A.1 Hybrid IO table Production processes Inv. Cons. Output Natural units MM MFF BM BP MP PP Mch Obj RM RP LF IN BS n.d. d. MM 1843 1843 ktons MFF 36,864 94,801 46,080 123,500 3520 1000 20,000 20,000 3456 1950 22,834 100,000 474,006 GJ BM 704 900 700 2304 ktons BP 200 500 600 1300 ktons MP 47,401 13,000 5000 5000 70,401 m 2 PP 10,000 10,000 20,000 m 2 Mch 63,730 36,270 100,000 units Obj 32,065 67,935 100,000 units RM 691 691 ktons RP 390 390 ktons LF 22 57 28 74 2 13 42 48 2 1 14 713 60 813 1889 ktons landfilled IN 22 100 73 96 104 395 ktons incinerated BS 184 474 230 130 70 20 100 100 69 39 10,000 11,417 Service units Note: Inv. investment, Cons. consumption, n.d. non-durables, d. durables, Note: for abbreviations of the production processes, see the next Table A.2. Table A.2 Monetary IO table Inputs/outputs Production processes Inv. Cons. Output MM MFF BM BP MP PP Mch Obj RM RP LF IN BS n.d. d. Mined metal (MM) 3686 3686 Mined fossil fuel (MFF) 737 1896 922 2470 70 20 400 400 69 39 457 2000 9480 Basic metal (BM) 2816 3600 2800 9216 Basic plastics (BP) 800 2000 2400 5200 Metal packaging (MP) 2370 650 250 250 3520 Plastic packaging (PP) 500 500 1000 Machines (Mch) 6373 3627 10,000 Objects (Obj) 3207 6793 10,000 Recycled metal (RM) 691 691 Recycled plastic (RP) 390 390 Landfilling (LF) 22 57 28 74 2 13 42 48 2 1 14 713 60 813 1889 Incineration (IN) 43 200 147 192 208 790 Business services (BS) 184 474 230 130 70 20 100 100 69 39 10,000 11,417 Primary inputs Labor 2051 2906 2864 968 535 140 2623 3048 402 179 1535 716 8806 26,773 Capital costs machines 461 1185 576 325 18 5 250 250 17 10 236 49 1427 4809 Capital costs objects 230 593 288 163 9 3 125 125 9 5 118 25 714 2405 Use of residuals Waste metal 123 123 Waste plastic 117 117 Waste metal packaging Waste plastic packaging Waste machines Waste objects Supply of residuals Waste metal 69 30 24 123 Waste plastic 13 60 44 117 Waste metal packaging Waste plastic packaging Waste machines (continued on next page)

392 ECOLOGICAL ECONOMICS 59 (2006) 375 393 Table A.2 (continued) Inputs/outputs Production processes Inv. Cons. Output MM MFF BM BP MP PP Mch Obj RM RP LF IN BS n.d. d. Waste objects Total 3686 9480 9216 5200 3520 1000 10,000 10,000 691 390 1889 790 11,417 Note: Inv. investment, Cons. consumption, n.d. non-durables, d. durables, Table A.3 Product prices Commodity/material Price per kiloton Price per natural unit Natural unit Mined metal (MM) 2.00 2.00 ktons Mined fossil fuel (MFF) 2.00 0.02 GJ Basic metal (BM) 4.00 4.00 ktons Basic plastics (BP) 4.00 4.00 ktons Metal packaging (MP) 5.00 0.05 m 2 Plastic packaging (PP) 5.00 0.05 m 2 Machines (Mch) 10.00 0.10 units Objects (Obj) 10.00 0.10 units Recycled metal (RM) 1.00 1.00 ktons Recycled plastic (RP) 1.00 1.00 ktons Landfilling (LF) N/A 1.00 ktons landfilled Incineration (IN) N/A 2.00 ktons incinerated Business services (BS) N/A 1.00 Service units Waste metal 0.50 0.50 ktons Waste plastic 0.50 0.50 ktons Waste metal packaging 0.50 0.50 ktons Waste plastic packaging 0.50 0.50 ktons Table A.4 Composition of raw materials and wastes Materials Si (%) H (%) C (%) O (%) Fe (%) Total (%) Composition Unmined ore 32 18 50 100 SiO 2 (50%) and Fe (50%) Unmined fossil fuel 60 6 34 100 SiO 2 (50%) and C 2 H 4 (50%) Oxygen 100 100 O 2 Waste metal 100 100 Fe Waste plastic 14 86 100 C 2 H 4 Waste metal packaging 100 100 Fe Waste plastic packaging 14 86 100 C 2 H 4 Waste machines 3 17 80 100 Mixed Waste objects 5 33 62 100 Mixed Note 1: Si: Silicon; H: Hydrogen; C: Carbon; O: Oxygen; and Fe: Iron. Note 2: Mixed indicates that the physical good is made of several substances. Table A.5 Composition of physical goods (excluding packaging) Physical commodity Si (%) H (%) C (%) O (%) Fe (%) Total (%) Composition Mined metal (MM) 100 100 Fe Mined fossil fuel (MFF) 14 86 100 C 2 H 4 Basic metal (BM) 100 100 Fe Basic plastics (BP) 14 86 100 C 2 H 4 Metal packaging (MP) 100 100 Fe Plastic packaging (PP) 14 86 100 C 2 H 4 Machines (Mch) 3 17 80 100 Mixed Objects (Obj) 5 33 62 100 Mixed Recycled metal (RM) 100 100 Fe Recycled plastic (RP) 14 86 100 C 2 H 4 REFERENCES Adriaanse, A., Bringezu, S., Moriguchi, Y., Rodenburg, E., Rogich, D., Schutz, H., 1997. Resource Flows: The Materials Basis of Industrial Economies. World Resources Institute, Washington DC. Ayres, R.U., Kneese, A.V., 1969. Production, consumption and externalities. American Economic Review 59, 282 297. Ayres, R.U., Ayres, L.A., Warr, B., 2005. Is the US economy dematerializing? Main indicators and drivers. In: van den Bergh, J.C.J.M., Janssen, M.A. (Eds.), Economics of Industrial Ecology: Materials, Structural Change, and Spatial Scales. MIT Press, Cambridge, MA.

ECOLOGICAL ECONOMICS 59 (2006) 375 393 393 Cumberland, J.H., 1966. A regional inter-industry model for analysis of development objectives. Papers of the Regional Science Association 17, 65 95. Daly, H.E., 1968. On economics of a life science. Journal of Political Economy 76 (3), 392 406. Dietzenbacher, E., 2005. Waste Treatment in Physical Input Output Analysis. Working PaperUniversity of Groningen, Groningen. Dietzenbacher, E., Giljum, S., Hubacek, K., Suh, S., 2005. Physical Input Output Analysis and Disposal to Nature. Working PaperUniversity of Groningen, Groningen. Giljum, S., Hubacek, K., 2001. International Trade, Material Flows and Land Use: Developing a Physical Trade Balance for the European Union. Interim Report IR-01-059IIASA, Laxenburg. Giljum, S., Hubacek, K., 2004. Approaches of physical input output analysis to estimate primary material inputs of production and consumption. Economic Systems Research 16 (3), 301 310. Giljum, S., Hubacek, K., Sun, L., 2004. Beyond the simple material balance: a reply to Sangwon Suh's note on physical input output analysis. Ecological Economics 48, 19 22. Gravgård-Pedersen, O., 1999. Physical Input Output Tables for Denmark. Products and Materials 1990, Air Emissions 1990 92. Statistics Denmark, Kopenhagen, ISBN 87-501-1076-4. Hansen, E., 1995. Environmental Prioritising of Industrial Products. Environmental project No. 281, The Danish Environmental Protection Agency (in Danish). Hoekstra, R., 2005. Economic Growth, Material Flows and the Environment: New Applications of Structural Decomposition Analysis and Physical Input Output Tables. Edward Elgar Publishers, Cheltenham, UK. Hoekstra, R., van den Bergh, J.C.J.M., 2004. Structural decomposition analysis of iron and steel and of plastics. In: van den Bergh, J.C.J.M., Janssen, M.A. (Eds.), Economics of Industrial Ecology: Materials, Structural Change, and Spatial Scales. MIT Press, Cambridge, MA. Hubacek, K., Giljum, S., 2003. Applying physical input output analysis to estimate land appropriation (ecological footprints) of international trade activities. Ecological Economics 44, 137 151. Isard, W., 1972. Ecological Economic Analysis for Regional Development. The Free Press, New York. Konijn, P.J.A. (1994). The make and use of commodities by industries: on the compilation of input output data from the National Accounts. PhD thesis, Universiteit Twente, Enschede, The Netherlands. Konijn, P.J.A., de Boer, S., van Dalen, J., 1995. Material Flows and Input Output Analysis: Methodological Description and Empirical Results. Notanr: 006-95-EIN.PNR/int BPA-nr: 698-95-EIN. PNR/int. Sector National Accounts, Statistics Netherlands, The Netherlands. Konijn, P.J.A., de Boer, S., van Dalen, J., 1997. Input output analysis of material flows with application to iron, steel and zinc. Structural Change and Economic Dynamics 8, 129 153. Kratterl, A., Kratena, K., 1990. Reale Input Output Tabelle und ökologischer Kreislauf. Physica-Verlag, Heidelberg. Leontief, W., 1970. Environmental repercussions and the economic structure: an input output approach. Review of Economics and Statistics 52 (3, August), 262 271. Leontief, W., Ford, D., 1972. Air pollution and economic structure: empirical results of input output computations. In: Brody, A., Carter, A.P. (Eds.), Input Output Techniques. American Elsevier, New York, pp. 9 30. Mäenpää, I., 2002. Physical input output tables of Finland 1995 solutions to some basic methodological problems. 14th International conference on input output techniques, Montreal, Canada. Mäenpää, I., Muukkonen, J., 2001. Physical input output in Finland: methods, preliminary results and tasks ahead. Conference on Economic Growth, Material Flows and Environmental Pressure, Stockholm, Sweden. Miller, R.E., Blair, P.D., 1985. Input Output Analysis. Foundations and Extentions, Prentice-Hall, Englewood-Cliffs, NJ. Nebbia, G., 2000. Contabilià monetaria e contabilità ambientale. Economia Pubblica 30 (6), 5 33. Stahmer, C., Kuhn, M., Braun, N., 1997. Physical Input Output Tables for Germany, 1990. Working Paper No. 2/1998/B/ 1German Federal Statistical Office. Statistische Bundesambt, 2001. Endbericht zum Projekt A Physical Input Output Table for Germany, 1995. Vertragsnummer 98/ 559/3040/B4/MM, Wiesbaden. Stone, R., 1961. Input Output and National Accounts. OECD, Paris. Suh, S., 2004a. A note on the calculus for physical input output analysis and its application to land appropriation of international trade activities. Ecological Economics 48 (1), 9 17. Suh, S., 2004b. Materials and energy flows in industry and ecosystem networks. Life-cycle assessment, input output analysis, material flow analysis, ecological network flow analysis, and their combinations for industrial ecology. PhD thesis, Institute of Environmental Studies (CML), University of Leiden, The Netherlands. UN (United Nations), 1968. System of National Accounts. UN (United Nations), 1993. System of National Accounts. UN (United Nations), 2003. System of Environmental and Economic Accounting. United Nations, New York. Victor, P.A., 1972. Pollution: Economy and Environment. George Allen and Unwin Ltd, London. Weisz, H., Duchin, F., in press. Physical and monetary input output analysis: What makes the difference? Ecological Economics. WRI (World Resources Institute), 2000. Weight of Nations: Material Outflows from Industrial Economies. World Resources Institute, Washington DC.