Dr. Fridolin Krausmann Institute of Social Ecology Schottenfeldgasse 25 1070 Wien, Austria firstname.lastname@example.org http://www.uni-klu.ac.at/socec/
The notion of a (socio-) metabolic transition has been used to describe fundamental changes in socioeconomic energy and material use during industrialization. During the last century, Japan developed from a largely agrarian economy to one of the world's leading industrial nations. It is one of the few industrial countries that has experienced prolonged dematerialization and recently has adopted a rigorous resource policy. This article investigates changes in Japan's metabolism during industrialization on the basis of a material flow account for the period from 1878 to 2005. It presents annual data for material extraction, trade, and domestic consumption by major material group and explores the relations among population growth, economic development, and material (and energy) use. During the observed period, the size of Japan's metabolism grew by a factor of 40, and the share of mineral and fossil materials in domestic material consumption (DMC) grew to more than 90%. Much of the growth in the Japanese metabolism was based on imported materials and occurred in only 20 years after World War II (WWII), when Japan rapidly built up large stocks of built infrastructure, developed heavy industry, and adopted patterns of mass production and consumption. The surge in material use came to an abrupt halt with the first oil crisis, however. Material use stabilized, and the economy eventually began to dematerialize. Although gross domestic product (GDP) grew much faster than material use, improvements in material intensity are a relatively recent phenomenon. Japan emerges as a role model for the metabolic transition but is also exceptional in many ways.
The notion of (socio-) metabolic transition has been introduced to describe fundamental changes in socioeconomic energy and material use during industrialization (Fischer-Kowalski and Haberl 2007; Haberl et al. 2011). It has been argued that the transition from an agrarian to an industrial society typically implies a multiplication of both metabolic rates (material and energy flows per capita and year) and population and, consequently, also metabolic scale (the overall size of extraction and trade flows). Not only the size of material and energy flows but also the composition of these flows changes, and a shift from biomass toward mineral and fossil materials occurs (Krausmann et al. 2008b). Researchers have also shown that growth in material use typically is accompanied by a considerable decline in resource intensity (resource use per unit of gross domestic product [GDP]; Krausmann et al. 2009). Research so far has focused on the energy side of the metabolic transition (Gales et al. 2007; Krausmann et al. 2008c; Warr et al. 2010), and only limited empirical evidence from material flow accounting (MFA) exists. Most MFA studies are limited to a few decades, and, at best, time series begin in the 1970s (e.g., Russi et al. 2008; Schandl and West 2010). These data sets allow for the analysis of a specific phase of industrialization that was marked by the oil price shocks of the 1970s and the subsequent deceleration of growth in energy and material use in most countries (Bringezu et al. 2004; Krausmann et al. 2008c). These studies thus miss important phases of industrial development and the metabolic transition, such as the decades after World War II (WWII) and the emergence of a society of mass production and mass consumption, during which tremendous increases in per capita resource use in industrial countries occurred (cf. Grübler 1998; Steffen et al. 2007). We know of only two comprehensive long-term material flow data sets consistent with current conceptual and methodological standards in MFA: Schandl and Schulz (2002) analyzed material use in the United Kingdom for the period from 1850 to 1998, and Krausmann and colleagues (2009) recently published a global time series of material extraction for the last century.1 Long-term data have also been published for selected material flows—for example, for the United States (Rogich and Matos 2002).
This article presents a time series of material flow data for the Japanese economy for the period from 1878 to 2005.2 With this study, we significantly extend the period for which material flows in the Japanese economy have been compiled so far: A first material flow account for Japan, covering the period from 1975 to 1993, has been included in the seminal multinational MFA study by the World Resources Institute (Adriaanse et al. 1997), and Japan's Ministry of the Environment (e.g., Ministry of the Environment 2007) has compiled annual time series data beginning in 1980.3
The time period observed in this article begins right after the Meiji restoration and covers Japan's transformation from an agrarian country to one of the leading economies in the world, then the so-called lost decade after the Japanese economic bubble burst in the late 1980s and economic recovery in the 1990s (Allen 1981; Hentschel 1986). It also includes important incidents at the global scale, such as the world economic crisis in the 1930s, World Wars I and II, and the oil price shocks following the initial incident in 1973.
After a brief description of methods and data sources, we present annual time series data for the period from 1870 to 2005 for extraction, imports, and exports of materials by main material group and aggregate material flow indicators, such as domestic material consumption and material intensity. On the basis of these data, we investigate the path of the metabolic transition in Japan. We identify changes in material input and use during Japan's industrialization and discuss the significance of socioeconomic factors underlying the observed changes in the physical economy of Japan. Finally, we explore changes in the material intensity of the Japanese economy and the issue of dematerialization.
Methods and Data
We followed the basic principles and current international standards of economy-wide MFA, as proposed, for example, by the European Statistical Office (Eurostat 2009; see also OECD 2008) to account for used extraction and direct imports and exports of materials. Unused extraction and indirect flows associated with imports and exports were not accounted for (see, e.g., Bringezu et al. 2009). We applied standard estimation procedures for the extraction of flows not reported in statistical sources and adapted them for long-term historical application. Our data are thus consistent with the current state-of-the-art methods of MFA and are comparable to other existing material flow accounts for Japan covering more recent periods (see the Data Quality and Reliability section of this article). Our database provides material flow data at a medium level of aggregation, discerning a maximum of 61 material groups. For this article, we use aggregate information and present data on the level of four main material groups: biomass, fossil energy carriers, metal ores, and nonmetallic minerals (including construction minerals). Data on fossil energy carriers, metal ores, and nonmetallic minerals are also subsumed under mineral and fossil materials (as opposed to biomass).
We calculated the following material flow indicators (see Fischer-Kowalski et al. 2011): domestic extraction (DE), imports and exports, domestic material consumption (DMC, defined as DE plus imports minus exports), physical trade balance (PTB, defined as imports minus exports), trade dependency (defined as PTB per DMC), and material intensity (MI, defined as DMC per unit of GDP; MI is the inverse of material productivity). We refer to per capita values of domestic material (and energy) consumption as metabolic rates.
We quantified material extraction and use for Japan proper as defined by the statistical sources we have used for the respective period. We did not include material extraction in Japanese colonies (e.g., Formosa [Taiwan], Chosen [Korea], or Manchuria), but trade between Japan proper and its colonies was, rather, accounted for as foreign trade. In practical terms, this means that the observed territorial system roughly resembles that of Japan in its current boundaries throughout the observed period.4 Our main data source was the excellent Japanese historical statistics database maintained by the Statistics Bureau Japan (2008), which provides, among other data, comprehensive physical information on agriculture, forestry, fishing, mining, the production of related industries, and trade. It contains annual time series data from as early as 1868 to the most recent years. Some of the series are shorter, however, and several cover the period after WWII only. Other important publications on Japanese historical statistics include a comprehensive data collection by the Bank of Japan (1966) and early data on foreign trade by Ishibashi (1935). Additionally, we used international data compilations and sources to complete and cross-check our data series. These data compilations include those from Mitchell (2003), FAOSTAT (2010), UN (2007), IEA (2007), USGS (2008), and United Nations Statistical Division (2008) and more specific literature. Data on population and GDP (in international Geary-Khamis Dollars) were taken from the work of Maddison (2008).5
We combined information on crop harvest and fish catch from Statistics Bureau Japan (2008) and FAOSTAT (2010). Complete and comparable data were available for the whole period. We estimated crop residues using region-specific information on corn-straw rations (harvest indexes) from the work of Krausmann and colleagues (2008a), for which we assumed changes over time (i.e., a 20% to 80% increase in harvest indexes during the last 130 years, depending on crop species). Grazed biomass was estimated on the basis of data on livestock numbers (Statistics Bureau Japan 2008; FAOSTAT 2010) and average feed intake per head and day. Feed intake data were calculated on the basis of ruminant production (milk yield per cow, average live weight) and assumptions on productivity changes. To arrive at the amount of grazed biomass and other roughage harvested to feed livestock, we subtracted information on available feed (market feed, fodder crops, and crop residues used as feed) from total demand. Data on feed supply were derived from the work of Statistics Bureau Japan (2008) and FAO (2010). For the period from 1878 to 1961, only limited information on the volume of available market feed was available, and we assumed that grazed biomass and harvest from grassland covered 70% of the feed demand of grazers. The feed balance was performed in dry matter; we converted results into MFA-relevant mass, assuming an average moisture content of 15% (Eurostat 2009).
Data from Statistics Bureau Japan (2008) cover fuel wood extraction beginning in 1929 and lumber harvest beginning in 1954. This information (which includes some statistical breaks and implausibilities, most likely due to unit confusions in the sources) was completed with data from the Bank of Japan (1966) and FAOSTAT (2010) and cross-checked with information from Mitchell (2003). For the period prior to 1929, fuel wood harvest was extrapolated from the average per capita consumption of the early 1930s and population numbers. We assume that official fuel wood data underestimate actual use in early periods, particularly in the period prior to WWII, for which we arrive at a comparatively low per capita DMC of fuel wood, 0.2 tonnes per capita per year (t/cap/yr). Information on the extraction of nontimber and fuel-wood products from forests was scarce; however, the involved mass flows are typically very low.
Data on mineral extraction provided by Statistics Bureau Japan (2008) begin in 1874; however, for many mineral materials, data series only begin in the early 20th century. To complement data for the years prior to WWII, we used data provided by the Bank of Japan (1966) and Torgasheff (1930). For the period from 1960 to the present, we also used data from USGS (2008) and the UN (2007) to cross-check and amend the database. We calculated the amount of extracted gross ores using information on metal content, ore grades, and coupled production derived from the work of USGS (2008) and Torgasheff (1930). The applied extrapolation coefficients were kept constant over time.
Data on construction minerals were not reported in available statistical sources, except for some years and some specific items. In particular, flows of natural aggregates used in construction and limestone for cement production had to be estimated. We used the procedure proposed by Krausmann and colleagues (2009) (and applied in a slightly modified form by Schandl and West  in a recent study on material flows in the Asia-Pacific region) and estimated the demand for sand and gravel used for concrete and asphalt production on the basis of data on cement and bitumen consumption. We assumed a ratio of sand and gravel to cement in concrete of 6.1 and of gravel to bitumen in asphalt of 20. Furthermore, we assumed that 1.15 tonnes of limestone are required to produce 1 tonne of cement (Krausmann et al. 2009). Data on cement production and consumption were derived from the work of Cembureau (1998), Statistics Bureau Japan (2008), and Bank of Japan (1966); bitumen consumption was derived from data in the work of IEA (2007) and Statistics Bureau Japan (2008). Additionally, we calculated sand and gravel demand for the construction of railroads assuming 10,000 t of sand and gravel per kilometer of newly built railroad tracks (Bank of Japan 1966; Statistics Bureau Japan 2008) to cover important construction activities in earlier periods. In general, the applied coefficients are conservative, and it can be assumed that the procedure has a tendency toward underestimating the overall amount of natural aggregate use: Our estimate emphasizes natural aggregates, which in most countries account for more than 90% of construction minerals, but we neglect other materials, such as clay for bricks. Also, filling materials are not fully accounted for (Krausmann et al. 2009).
For the period from 1960−1961 to 2005, we used data from FAO (2010) for trade with agricultural and forestry products, data from IEA (2007) for trade with fossil energy carriers, and trade data from United Nations Statistical Division (2008) at the three-digit level of the Standard International Trade Classification (SITC, Revision 1; approximately 300 items) for trade with all other products. Data for the period from 1946 to 1960 were derived from Statistics Bureau Japan (2008) and are presumably incomplete (we assume that we do not account for some 20% to 40% of the total trade flows in mass in that period). Data for the period 1870–1933 are based on the work of Ishibashi (1935) and include trade with Japanese colonies. Trade between Japan proper and its colonies was significant and accounted for 23% of total imports and 26% of total exports in 1933. For the period from 1934 to 1945, no trade data were available. Although trade flows in that period might have increased, we assume that the overall mass of trade flows remained small compared to DE; after WWII, trade volumes (in monetary terms) reached the level of the mid-1930s only in the late 1950s (Allen 1981).
Data on material flows were used to calculate total primary energy supply (TPES). We converted fuel wood and fossil fuel DMC into energy units using material-specific coefficients for gross calorific values. Energy flow data derived from MFA were supplemented with energy inputs from hydropower, nuclear heat, and geothermal sources. We converted data on electricity output from hydro and nuclear power plants into primary energy input by applying coefficients for average conversion efficiency (Warr et al. 2010).6
Data Quality and Reliability
One of the reasons that most MFA studies rarely extend earlier than the 1970s is that compiling long-term time series data of sufficient robustness and comparability is a difficult and laborious process. International data compilations and digitally available data get increasingly scarce for periods prior to the 1970s. In the case of Japan, we could compile a time series covering 135 years because the country has a long tradition of statistical records and comprehensive sources for historical statistics (e.g., Statistics Bureau Japan  provided a reliable empirical backbone for the material flow account). Long-term data do have their weaknesses, however, and uncertainty tends to increase the further time series are extended to historic periods, as underreporting, data gaps, and flaws become more frequent.
Japan has a long tradition in statistics, and the quality of the used sources must be regarded as very high. Nevertheless, we assume that we slightly underestimate material extraction, in particular for the early years. As has been outlined in the Methods and Data section, our estimates for construction minerals and some biomass flows, in particular the extraction of wood and other forest products, have to be considered conservative. Furthermore, the data coverage is insufficient for the period from 1934 to 1945, for which only data on DE, but no trade data, were available. Also, for the years 1946 to 1960, trade data were incomplete, and net imports have to be considered to be low (roughly by 20% to 40%). Hence, aggregate indicators for this period must be interpreted cautiously. Despite these caveats, it can be assumed that our data represent the size of material flows for the four material groups and their development over time fairly well, even in the time period from 1878 to 1960. We assume that possible underestimations of some flows are not significant enough to distort the overall picture of trends in material use over time.
The reliability of our data is further corroborated when compared with official Japanese MFA data: For the period from 1980 to 2004, MFA accounts published by the Japanese Ministry of the Environment (2007) are available. For most flows and material groups, our data match very well with the Japanese data set; this is in particular encouraging for domestic extraction, for which large material flows were estimated (e.g., grazed biomass, sand and gravel, and gross ores). For the extraction of gross ores (which is a very small flow compared to the other main material groups in Japan) we arrive at considerably higher figures than those reported in the official Japanese data set; however, an in-depth review showed that Japanese data most likely refer to metal content rather than gross ore. Figure 1 shows that our results are very similar to official Japanese MFA data with respect to both the overall amount of DMC and its development over time. In terms of trends over time, our data also match well with the Japan data published recently in a multinational material flow database for the Asia-Pacific covering the period from 1970 to 2005 (Schandl and West 2010). Schandl and West's data are, however, not based on national statistical sources or country-specific coefficients used in estimates, which explains differences in metabolic scale (figure 1).
Domestic Extraction, Trade, and Domestic Material Consumption
Figure 2 shows the development of DE, trade flows, and DMC by the four main material groups. In 1878, biomass accounted for almost 90% of DE, but the extraction of fossil energy carriers and mineral materials increased continuously, and by 1923 it surpassed biomass in terms of mass. Between 1878 and the 1940s, total extraction multiplied almost sevenfold, from 0.03 Gigatonnes (Gt; billion tonnes) to 0.2 Gt. At the end of WWII, DE slumped to only half of the prewar value, in particular the extraction of mineral materials. After 1947 DE quickly recovered; it had already surpassed the prewar peak by 1951 and then experienced a sheer explosion in the period up until 1973. The steep increase in overall DE was caused by a surge in nonmetallic minerals, mostly natural aggregates for construction, which dwarfed all other material flows. The steep rise of nonmetallic minerals DE was interrupted by the oil price shock in 1973, but DE continued to rise, to reach its maximum in 1991. The development of the extraction of all material groups shows an inverted U shape and declines after early peaks: DE of fossil energy carriers peaked as early as 1943, not quite reaching this level again in a second peak after WWII in 1962. DE of biomass peaked in 1960, and DE of ores peaked in 1967. During the observed period, aggregate extraction grew 36-fold and reached a peak in 1991 at 1.4 Gt (i.e., 3% of global DE at that time). Biomass was the material group with the lowest growth; it quadrupled between 1878 and 1960. After 1960, DE of biomass declined from 0.13 to 0.08 Gt, or by 40%.
Trade flows exhibit an even more extreme growth pattern (figure 2). Mass flows entering or leaving Japan through trade were very small in the decades before WWII: Total imports only reached one-tenth of DE in the 1930s, and exports never exceeded 5% of DE. The most relevant trade flows in this period were imports of biomass, above all food items and cotton, and, in later years, also of ores. On the export side, coal and, increasingly, manufactured products were the dominating mass flows. Unfortunately, no trade data for the period from 1934 to 1955 are available, although it can be assumed that imports of wood and minerals from Manchuria and other colonies were considerable. After the war, both imports and exports experienced steep growth: They grew at average annual rates of 21% and 26%, respectively, during the period from 1955 to 1971. The surge of imports was interrupted in 1973 (due to the oil price shock) but continued in the 1980s, whereas the growth of exports slumped between 1985 and 1990. By 2005, exports had reached 15% of DE and imports more than 80%. During the whole period after WWII, imports were much larger than exports, and Japan remained a massive net importer of all four material groups. The country imported large amounts of biomass and fossil energy carriers but also ores and minerals. Exports mostly consisted of products from metals and nonmetallic minerals.
As a result of the development of DE and trade flows, DMC (figure 2D) grew slowly in the period before WWII, from 0.04 Gt in 1878 to 0.23 Gt in the 1940s, but surged in the 1950s and 1960s, to 1.8 Gt in 1973. The effect of the oil price shocks on DE and imports is clearly reflected in material consumption, which dropped considerably in 1973. After 1973, the previous growth dynamic came to a halt. DMC experienced major ups and downs and reached its peak at 2 Gt in 1990. Ever since, Japan's economy has been dematerializing, and DMC decreased by 15%, to 1.7 Gt in 2005.
The Dynamics of Growth
Japan presents itself as a showcase of the metabolic transition from an agrarian to an industrial metabolic regime. During the observed 130-year period, the country's population grew by a factor of four, but metabolic rates surged by more than an order of magnitude: Material use (DMC/cap/yr) grew by a factor of 14, and primary energy supply (TPES/cap/yr) grew by a factor of 50 (table 1 and figure 3). Although biomass dominated material and energy use in the late 19th century, its contribution to DMC and TPES was dwarfed to less than 10% a century later (table 2). This transition was by no means a steady process. Three periods with distinct patterns of change in material and energy use can be identified: a period of moderate physical growth from 1878 to the outbreak of WWII (interrupted by the world economic crises of the 1930s); a period of radical growth beginning shortly after WWII and lasting until the early 1970s; and, finally, a period of stagnation beginning in 1973, which was characterized by strong fluctuations and, eventually, dematerialization. Table 1 shows average annual growth rates for major physical and economic headline indicators for these three periods. Growth rates for all resource use indicators are moderate in the first phase, increase by a factor of three to five during the second phase, and are low or even negative during the third phase.
Table 1. Average annual growth rates of population, gross domestic product (GDP), and material and energy use during periods of development
Moderate growtha 1878–1930 (%)
Rapid growth 1948–1973 (%)
Stagnation 1973–2005 (%)
Growth factor 2005/1878
Note: intl. $= international Geary-Khamis $; DMC = domestic material consumption; TPES = total primary energy supply. One tonne (t) = 103 kilograms (kg, SI) ≈ 1.102 short tons; one gigajoule (GJ) = 109 joules (J, SI) ≈ 2.39 × 105 kilocalories (kcal) ≈ 9.48 x 105 British Thermal Units (BTU).
Source: Our own calculations.
aWe calculated growth rates only for the period from 1878 to 1930 to avoid distortions from the world economic crisis and incomplete data (see Methods and Data section).
GDP (intl. $)
DMC minerals/fossils in t/cap/yr
TPES in GJ/cap/yr
Table 2. Japan's metabolic transition: Key indicators for selected years
Note: DMC = domestic material consumption; GDP = gross domestic product; PTB = physical trade balance; TPES = total primary energy supply; DEC = domestic energy consumption; intl. $= international Geary-Khamis $. One megawatt-hour (MWh) ≈ 3.6 × 109 joules (J, SI) ≈ 3.412 × 106 British Thermal Units (BTU); one gigajoule (GJ) = 109 joules (J, SI) ≈ 2.39 × 105 kilocalories (kcal) ≈ 9.48 x 105 British Thermal Units (BTU); one hectare (ha) = 0.01 square kilometers (km2, SI) ≈ 0.00386 square miles ≈ 2.47 acres.
Source: Rural population (Allen 1981); population density and income based on data from Maddison (2008); all others are from own calculations based on the material flow accounting (MFA) database.
Share of agricultural population
% of total
DMC per capita
PTB as % of DMC
Share of biomass in DMC
% of total DMC
TPES per capita
Share of biomass in TPES
% of total DMC
Electricity per capita
Power density (DEC per unit area)
Material load (DMC per unit area)
In the 1880s, during the Meiji period, the metabolism of the Japanese economy resembled the metabolic profile of a typical agrarian regime (Krausmann et al. 2008b). Four-fifths of the population was making its living in agriculture (table 2); material consumption amounted to merely 1.1 t/cap/yr, and biomass accounted for almost 90% of DMC. Nearly all material resources were met from domestic extraction; imports and exports were insignificant. But the shift from an organic to a mineral economy (Wrigley 1988) was already underway. The so-called Meiji restoration at the end of the 1860s led to the abolishment of feudal traditions and institutions and opened the country to the West. Japan began to industrialize and engaged in international trade. Much of this early industrialization was in the silk and textile industry in small-scale factories and based on biomass as the primary raw material. The expansion of metal manufacturing and chemical industries in the late 1930s resulted in a more material-intensive development and an increase in the per capita consumption of mineral and fossil materials (figure 3B). In this period, the merchant fleet and the railway system were expanded, and exports in monetary terms surged (Allen 1969). Japan's (military) expansion in the Far East, which culminated in the Sino-Japanese War in 1937, significantly contributed to rapid industrialization in this period (Allen 1981). The corresponding material flows, however, remained comparatively low (figure 2). Japan was exporting considerable amounts of ores and, later, also coal and importing predominantly food and cotton. But overall, imperial Japan rather followed a policy of economic self-sufficiency and trade dependency (net imports as a share of DMC remained low, at only 8% in 1930; figure 4). In the first decades of the 20th century, heavy industry also gained significance and contributed to physical growth, but its per capita output in physical terms remained rather low.7 Industrial growth is reflected in increasing but, overall, still low values of material and energy use per capita (figure 3). But even in this period of moderate physical growth, material and energy use increased faster than population and GDP (table 1). All in all, per capita material use tripled, and the share of biomass in DMC declined to less than 50% in the years prior to Japan's involvement in WWII.
WWII left Japan with massive destruction. Most cities, industrial buildings, and plants were devastated.8 Raw materials were scarce, and food and other necessities of life were lacking (Allen 1981). In the 1950s, pushed by economic, social, and institutional reforms of the occupational authorities and demand increases induced by the Korean War, Japan's economy recovered quickly. In the 2 decades that followed Japan's independence in 1952, the Japanese economy grew faster than any other large economy in the world (Allen 1981). This period saw the building of great trunk roads and elevated highways, the modernization of the rail system, and a massive expansion of heavy industry and manufacturing.9 Concrete rapidly replaced the traditional building material of wood, and stocks of built infrastructures increased to several hundred tonnes per capita (Hashimoto et al. 2007; Tanikawa and Hashimoto 2009). Japan experienced a consumption revolution, and material benefits of economic development now also reached rural areas. This triggered rapid physical growth and radical changes in Japan's metabolism. Material and energy use grew much faster than population and even faster than GDP. Per capita material and energy use saw average annual growth rates of 8.5% and 12%, respectively (see table 1 and figure 3), and the share of biomass in DMC declined rapidly (figure 2D). Under U.S. occupation, Japan dismissed its autarky policy and sought integration into international markets. Japan began to import most raw materials and energy carriers. Within less than 20 years, trade dependency for fossil energy carriers and ores surged from almost zero to more than 90% (figure 4). Domestic extraction of these materials, in contrast, peaked and began to decline in the 1960s (see the Domestic Extraction, Trade, and Domestic Material Consumption section of this article). In 1973, material consumption reached an all-time high of 16.5 t/cap/yr, which was six times the pre-WWII level. The share of biomass had declined to less than 10%.
Things changed in 1973. Japan, which was depending massively on oil imports for its industrial development, felt the oil price shock sharply; the drastic increase in industrial costs cut short the boom. Although income (GDP/cap/yr) quickly recovered after 1973, the impact of this event on the physical economy was longer lasting. After a period of strong fluctuations in the 1970s, per capita DMC began a continuous decline following the economic crisis of the 1990s (figure 3). In 2005 it reached 13.3 t/cap/yr, or the level of the early 1970s. Energy use, however, showed a different pattern: TPES recovered in the mid-1980s and grew significantly until the late 1990s, but the composition of energy use changed. Per capita consumption of petroleum declined, and energy provision shifted toward coal and nuclear power. The import dependency of the Japanese economy was not lastingly affected, either. Even imports of fossil energy carriers soon began to rise again (figure 4). In 2005, import dependency amounted to almost 100% for ores and fossil energy carriers and 55% for biomass—higher values than those observed for the economies of Belgium and the Netherlands, which have the highest trade dependency in the European Union (Weisz et al. 2006).
Material Intensity and Dematerialization
Since the late 19th century, the Japanese economy has experienced tremendous growth, in both physical and monetary terms. This section explores the links between growth in GDP and material use. Between 1878 and 2005, GDP grew significantly faster than DMC, and aggregate material intensity declined by more than 50%, to 0.6 kg/$ in 2005. Figure 5A shows that aggregate material intensity remained at a high level until the early 1970s, oscillating between 1.2 and 1.7 kg/$. Only after the first oil price shock (1973) did a rapid decline of material intensity set in. The contribution of individual material groups to this aggregate trend differed largely: Although the material intensity of biomass DMC has declined continuously since the late 19th century, the use of mineral and fossil materials grew much faster than GDP. Material intensity for this group grew from 0.1 kg/$ in 1880 to 1.3 kg/$ in 1974. Only then did the trend reverse itself, and the following sharp decline characterized the development of aggregate material intensity in the period from 1974 to 2005.
The case of Japan shows that significant increases in material productivity (i.e., a declining material intensity) are a comparatively recent phenomenon. Biomass is a resource that is scarcely related to economic development and much more linked to population growth (Steinberger et al. 2010), whereas mineral and fossil materials, key resources of industrial development, seem to grow faster than GDP throughout large periods of industrialization. Similar long-term trends have also been observed for material intensity at the global scale (Krausmann et al. 2009) and for the ratio of useful work and GDP in several industrialized countries, including Japan (Warr et al. 2010).
Since the 1970s (when empirical evidence becomes available), most industrialized countries have shown increases in material productivity. But any improvements are usually outgrown by GDP, and, despite a considerable decline in material intensity, only relative decoupling of economic growth and material use is achieved (Bringezu et al. 2004; Steger and Bleischwitz 2009). Japan is one of the rare exceptions showing a significant and long-lasting decline in aggregate material use—or, in other words, absolute dematerialization.10 After a period of strong fluctuations following the oil price shocks of the 1970s, aggregate DMC peaked in 1991 (figure 2D). After that time, a lasting decline began that resulted in a 17% reduction by 2005.11 Energy use showed a different development. TPES per capita did recover in the 1980s and continued to rise, with changes in composition (see figure 3). Our data show that much of the observed decline in DMC was due to reduced consumption of construction minerals, although DMC of biomass and ores declined also. In contrast, the consumption of fossil energy carriers continued to increase, contributing to the observed growth in TPES. The reductions in material use can predominantly be attributed to reductions in domestic extraction, whereas imports of most materials continued to grow (figure 2). Hashimoto and colleagues (2008) found that in the period from 1995 to 2002, changes in the demand structure—above all, a shift from construction toward machinery and services—contributed the largest share of improvements in material productivity and dematerialization (in spite of an increase in material input per unit of constructed buildings and infrastructures due to more restrictive building codes that were implemented after the Kobe earthquake of 1995; see Tanikawa and Hashimoto 2009). Which activities or measures were actually driving the remarkable decline in material use is still largely obscure and requires further research.
Figure 5B shows annual growth rates of GDP and DMC and exhibits a high correlation between economic and physical growth. The oil price shock of 1973 obviously brought an abrupt end to high economic growth rates, which from then on rarely exceeded 4% per year. DMC very much followed the trend of GDP; growth rates of DMC declined markedly after 1973 and even became negative in many years. We see that the decline in growth rates of DMC in Japan coincides with a long-lasting reduction of GDP growth, which corroborates previous findings that a stabilization or a decline of DMC in many cases only occurred during periods of low economic growth or recession (Mudgal et al. 2010) and vice versa: High economic growth usually goes hand in hand with physical growth. Since 2000, Japan has officially pursued the goal of establishing a sound material cycle society (Junkangata Shakai) and reducing material consumption by following the so-called 3R policy (reduction, reuse, and recycling). Japan is one of very few countries that has adopted a resource policy with clear targets aiming at an overall reduction of material use (Takiguchi and Takemoto 2008). It is generally assumed that this policy has further enhanced a trend of absolute dematerialization in Japan, which has its roots in economic restructuring after the Asian economic crisis (see, e.g., Hashimoto et al. 2008). Detailed knowledge about the actual effect of the Japanese resource policy and its specific measures on the observed trends is still lacking, however.
The Metabolic Transition in Japan: Role Model or Special Case?
Japan was a late comer to the Industrial Revolution, but it caught up with the leading economies at an unprecedented speed. During Japan's development from a largely agrarian economy to one of the world's leading industrial nations, its material turnover grew 49-fold, and the share of mineral and fossil materials in DMC surged to more than 90% (table 2). Despite considerable population growth, growth in DMC was driven by increases in metabolic rates rather than by the growing number of inhabitants. Over the whole period, the 14-fold increase in per capita material use accounted for three-quarters of the total increase in DMC. These profound changes in size and composition of material use are typical for the metabolic transition from an agrarian to an industrial metabolic regime (Fischer-Kowalski et al. 2007).
But the Japanese case appears exceptional in several ways: For instance, the speed of the transition is remarkable. After WWII, Japan's metabolism experienced sheer explosive growth, and a considerable part of per capita growth in material and energy use occurred during a time span of merely 20 years, during which Japan developed a massive heavy industry, built up large physical stocks in buildings and infrastructure, and adopted mass production and consumption. But in the 1970s, coinciding with the first oil price shock, growth came to an abrupt halt.
Another remarkable feature is Japan's high dependency on trade: Very early on, the metabolic transition in Japan was based on imported raw materials rather than on domestic resources. As a result, the Japanese physical economy today depends to a degree on net imports, which is usually only seen in small countries with a very specialized economy.
Finally, the fact that Japan experienced absolute dematerialization over an extended period (not only during a recession or induced by massive deindustrialization) deserves special attention: Japan's DMC peaked at 16.5 t/cap/yr in 1973 and in 2005 was as low as 13 t/cap/yr. This is far below average material use in the OECD (20t/cap/yr) and in the European Union (15 t/cap/yr) and one of the lowest of all high-income countries.12 But caveats are warranted: The high significance of net imports, which amount to more than 5 t/cap/yr, or almost 50% of DMC, may be one important cause for low metabolic rates and low material intensity. Net importing countries externalize a good part of material-intensive production processes, and several studies conclude that high import dependency can contribute to high embodied consumption and emissions (see, e.g., Stromman et al. 2009; Hertwich and Peters 2009 for carbon dioxide [CO2]). For countries such as Austria, Germany, and the Czech Republic (import dependency of less than 20%), it has been shown that DMC would be roughly 30% higher if upstream flows of net imports were accounted for (see, e.g., Mudgal et al. 2010).
The study of long-term trends in socioeconomic material use helps researchers to better understand the dimensions of changes in social metabolism that occur with industrialization and highlights the significance of the metabolic transition for sustainable development. The case of Japan shows very clearly that although severe (and successful) efforts have been made to reduce materials use and increase resource efficiency, laying a path toward sustainable levels of resource use still remains a major political challenge.
This work was supported through the GLOMETRA project, funded by the Austrian Science Fund (FWF; No. P-21012-G11). We thank Julia Steinberger for commenting on previous versions of the article, Hiroki Tanikawa for his help with Japanese sources, and three anonymous reviewers for their critical comments.
As this article was going to press, we identified a recently published study on long term material flows in Czechoslovakia by Kovanda and Hak (2011).
See also the work of Moriguchi (2001, 2002) for a discussion of early material flow data of the Japanese economy.
Minor deviations from these territorial system boundaries are possible; often, sources are not explicit about their reference system, and in some cases not all prefectures or islands have been included in the surveys (see Statistics Bureau of Japan 2008).
The Geary-Khamis dollar, more commonly known as the international dollar, is a hypothetical unit of currency that has the same purchasing power that the U.S. dollar had in the United States at a given point in time. All currency used throughout the article is in Geary-Khamis Dollars.
Although most of the electricity produced in Japan comes from thermal power plants, this is already accounted for in the consumption of fossil energy carriers. In contrast, the primary power and heat used to produce electricity in hydro and nuclear power needs to be extrapolated from electricity output (Haberl 2001).
Per capita output of Japanese heavy industries remained low: Steel production, for example, barely reached 100 kg/cap/yr in the years prior to WWII, compared to around 350 kg/cap/yr in Germany or the United States (on the basis of Mitchell's  work). Similarly, Japanese cement production (per capita) amounted to less than half the value typical for European industrializing countries (Cembureau 1998).
The amount of physical destruction has been estimated at twice the national income of the fiscal year of 1948−1949 (Allen 1981).
Per capita steel output of the Japanese economy surpassed that of the United States, the United Kingdom, and Germany in the 1960s and peaked at 1,100 kg/cap/yr in 1974. At 700 to 800 kg/cap/yr at the beginning of the 21st century, it was still higher than in most industrialized countries.
In contrast to absolute dematerialization, relative dematerialization occurs when DMC is growing, but at a slower pace than GDP. As a consequence, material intensity of the economy is declining, whereas material use continues to rise.
According to the official Japanese MFA data (Ministry of the Environment 2007), the peak in material use was already reached in 1990, and from then on DMC declined by 29% until 2004.
The United Kingdom and Switzerland are the only fully industrialized countries that have a similarly low per capita DMC.
About the Authors
Fridolin Krausmann is a professor of sustainable resource use and deputy director of the Institute for Social Ecology at Alpen Adria Univeritaet in Vienna, Austria. Simone Gingrich is a research associate at the Institute for Social Ecology at Alpen Adria Univeritaet. Reza Nourbakhch-Sabet is a researcher at the Institute of Social Ecology at Alpen Adria Universitaet.