• Open Access

From Energy to Environmental Analysis

Improving the Resolution of the Environmental Impact of Dutch Private Consumption with Hybrid Analysis


  • René M.J. Benders,

  • Henri C. Moll,

  • Durk S. Nijdam

René M. J. Benders, Center for Energy and Environmental Studies (IVEM), University of Groningen, Nijenborgh 4, 9747AG, the Netherlands. Email:r.m.j.benders@rug.nl


Unsustainable private consumption causes energy and environmental problems. This occurs directly (resource depletion and emissions through using cars for transport) or indirectly (purchase of consumer goods and services for which the production uses energy and emits damaging gases).

A hybrid energy analysis proved that indoor energy consumption, mobility, and vacations are the main consumer categories from an energy point of view. Although energy is often used as a proxy for environmental load from private consumption, there are other proxies like methane (CH4), sulfur oxides (SOx), and land use. This article describes the results of the extension of the hybrid energy analysis with energy and ten environmental stressors (CH4, nitrous oxide [N2O], nitrogen, phosphate, SOx, nitrogen oxides [NOx], ammonia [NH3], nonmethane volatile organic compounds [NMVOCs], particulate matter [PM10], and land use), combined in five impact categories (global warming potential [GWP], acidification, eutrophication, summer smog, and land use).

Household consumption was analyzed by dividing Dutch household expenditure into 368 consumer items in 11 categories. The results show that food impacts, in particular, are underestimated when only energy is taken into account. Food makes the highest contribution in three out of five impact categories when all ten stressors are taken into account. Within the food domain, meat and dairy consumer items have the highest environmental impact, about 45% of total food impact on average across all five impact categories. Looking in detail (368 consumer items), there are nine food items in the top ten most-polluting items. Salad oil and cheese are the most polluting food items.


“The major cause of the continued deterioration of the global environment is the unsustainable pattern of consumption and production, particularly in industrialized countries.” This was stated in Agenda 21, chapter 4.3 (UN 1992) and was reaffirmed at the World Summit on Sustainable Development in Johannesburg, South Africa, in 2002 (UN 2002). The concept of sustainable consumption involves both sustainable production by producers and the sustainable behavior of consumers. In the past, environmental policy was mainly directed at the supply side (producers); currently, the demand side (consumers) is being given full attention by policymakers. It is widely recognized that changing household consumption patterns is essential for achieving sustainable development.

Consumption has played a central role in economic growth since the start of the Industrial Revolution. Private consumption per capita has increased steadily and is expected to continue to follow gross domestic product (GDP) growth in the next two decades (OECD 2002). Consumer activities can be linked to economic activities because consumption leads to production, a pattern that is “the major cause of continued deterioration of the global environment” (UN 1992). A major part of consumer activity is determined in households; therefore, most of the environmental load in an economy can be allocated to households (Biesiot and Noorman 1999).

In this research, we focus on the demand side, and specifically on the environmental load of households. Different methods can be used to calculate the environmental impact of household expenditure, such as process analysis and input-output (IO) analysis. A cradle-to-grave approach characterizes both methods.

Energy and/or environmental analyses of consumption have been performed for several countries. Interesting publications in this field include the calculation of direct and indirect energy requirements of households for Brazil (Cohen et al. 2005), India (Pachauri and Spreng 2002), the Netherlands (Vringer and Blok 1995), and a comparison between four countries (the United Kingdom, Sweden, Norway, and the Netherlands; Moll et al. 2005). Munksgaard and colleagues (2000) calculated the carbon dioxide (CO2) emissions of Danish households, Alfredson (2004) calculated energy requirements and CO2 emissions of Swedish households, and Carlsson-Kanyama and colleagues (2005) calculated household energy consumption for several types of households in Stockholm. As environmental statistics improved, the range of research broadened further than energy and/or CO2 emissions. Lenzen (1998) studied greenhouse gases (CO2, CH4, and N2O) for Australian final consumption. More recent research has been published by Nijdam and colleagues (2005), who calculated several environmental indicators for the Netherlands; Munksgaard and colleagues (2005), who performed a similar study for Denmark, and Tukker and colleagues (2005), who performed a study for the 25 European Union countries (EU-25). Weber and Matthews (2008b) did an analysis of global aspects of CO2 emissions in the United States. All of the studies mentioned above are IO based. The exceptions are those of Vringer and Blok (1995) and Alfredson (2004), which use a mix of both IO and process-based analysis. This mix is the so-called hybrid method, developed by Bullard and colleagues (1976, 1978) and worked out in a software package by Wilting, Kok, and Benders (Wilting 1996; Wilting et al. 1999; Benders et al. 2001; Kok et al. 2006).

Energy use is often considered to be a proxy for environmental load according to the articles cited in the previous paragraph. For climate change and acidification, energy-related emissions are the major contributors. Energy use takes place in every sector of the economy and society (Huijbregts et al. 2006), while other inputs and outputs such as materials, land, and specific emissions are largely confined to one or a few specific sectors (ammonia [NH3] in agriculture and the fertilizer industry, building materials in housing, land use in agriculture and forestry). However, for a more integrated environmental evaluation of household consumption, other environmental stressors and/or impact categories should also be included. The omission of these other stressors will result in an incomplete picture of the environmental load of consumption. A multistressor approach is needed to overcome this deficiency.

In the approach followed in this article, stressors have been added and the results of this multistressor approach are compared with the single (energy) approach. The aim is to gain insight into the change in environmental evaluation using a more integrated method. To this end, the hybrid analysis method is expanded from a single to a multistressor approach with five impact categories: global warming potential (GWP; CO2, methane [CH4], nitrous oxide [N2O]); eutrophication (nitrogen, phosphate, nitrogen oxides [NOx], NH3); acidification (sulfur oxides [SOx], NOx, NH3); summer smog (NOx, nonmethane volatile organic compounds [NMVOCs]); and land use. With this extended method we calculated the environmental impact for total household consumption (12 COICOP1 domains), for food in more detail (11 COICOP subdomains), and for all 368 household expenditure items.

Using the hybrid approach in combination with the multistressor approach increases the level of detail/information for household expenditures. The multistressor approach can give better insights into decision support and options for change in a more environmentally beneficial direction, as promoted by Tukker and colleagues (2010). By using the hybrid method, this can be done better on a micro level so that decision support can even extend to the point stipulated by Tukker and colleagues (2006).


To calculate the indirect environmental load of an expenditure, the entire life cycle load (the so-called from-cradle-to-grave approach) should be calculated. This can be calculated in the same way as indirect energy requirements. To calculate the entire energy of the life cycle of a product, we used the hybrid method developed by Bullard and colleagues (1976, 1978). This is characterized as a bottom-up approach with a high level of detail (Kok et al. 2006). This hybrid method is a combination of IO analysis and process analysis,2 taking the best from both methods (Bullard and Pilati 1976; Bullard et al. 1978). As a consequence, the processes of a product life cycle are partly described in physical terms and partly in financial terms. (See figure 1, which presents the scheme for the hybrid method.3)

Figure 1.

Chart of the hybrid method for the life cycle calculation of environmental parameters of budget spending. The shapes indicate whether a calculation is based on process or input-output (IO) analysis.

The basic goods, packaging materials, transport, direct consumption in the households, and waste processing are described in physical terms by means of process analysis. These descriptions are linked to environmental data. Capital and residual goods, manufacturing, and trade are described in economic terms by means of IO analysis data and are also linked to environmental data. For all calculations, the assumption is made that imports are produced with the same technology and environmental load as in the Netherlands (see Discussion section of this article). Below, an example is given to illustrate the analysis of a consumer item.

In table 1, the GWP of the analysis of whole wheat bread is given and described in accordance with the flowchart in figure 1. The second column shows the amounts in physical or monetary units in the entire chain to produce 1,000 kg of whole wheat bread and 4.7 kg of packaging material. The third column gives the GWP intensity of the items in the first column. The last column is the environmental load, or the product of columns 2 and 3.

Table 1. Example analysis: Entire life cycle of global warming potential (GWP) emissions in carbon dioxide equivalents (CO2-eq) for the production of 1,000 kilograms (kg) of whole wheat bread
Expenditure Whole wheat bread
Manufacturer sector Bread and rusk factories, bakeries, etc.
Price incl. 6% tax (€) 1,557
Amount, unit 1,000 kg
Transport weight (kg) 1,004.70
Load Source Load Intensity Global Warming Potential
  1. Notes: incl. = including; €= euro; LDPE = low-density polyethylene; t = metric ton; km = kilometer; n.e.c. = not elsewhere classified.

  2. a Because the basic goods oil and flour contain short-cycle CO2 (biomass) and municipal residual waste is incinerated, the GWP intensity results in a negative value.

Basic goods kg GWP/kg GWP
Vegetable oil/fats30.000.87226.153
Wheat flour (whole wheat)700.000.611427.573
Packaging materials kg GWP/kg GWP
Manufacturer GWP/€ GWP
Capital goods32.020.52116.679
Residual goods347.761.001348.156
Transport km GWP/t/km GWP
Lorry (truck)100.000.22022.191
Trade/services GWP/€ GWP
Retail trade: groceries (general)381.980.327124.995
Wholesale trade: groceries n.e.c.197.870.25851.038
Waste processing kg GWP/kg GWP
Average residual waste (incl. transport)a25.25−0.266−6.722
Average plastics (incl. transport)4.200.8253.484
Global warming potential CO2-eq GWP/€ GWP/kg

Starting with the consumer price (1,557 euros [€]), the price for the trade sector is calculated by subtracting the value-added tax (VAT; 6%). By selecting the retail and wholesale sectors, the margins are subtracted from the remaining price and the accompanying environmental load for the trade sector is determined. The remaining €889.32 is the producer price, and is used to determine the environmental load of the manufacturer. Here the costs for the basic goods and packaging materials are subtracted, as are the energy costs, the value added, and the depreciation costs for the capital goods. The remaining amount is assumed to be needed to pay for the residual goods (€347.76). Residual goods are those materials present in the product but not defined as basic goods because they are not known or their share in the total weight is relatively small. Residual goods also include products and services that are indirectly used in production, such as office supplies. The environmental load of the residual goods (cumulative environmental intensities) is calculated with the aid of the IO table (with 113 sectors) and sector emissions. To avoid double counting, the sectors that produce the basic goods and packaging materials are excluded when calculating the cumulative environmental intensities. The environmental intensities for the production and the depreciation are available for each manufacturer. Although the use phase is present in the methodology, it is not used explicitly in this study. The energy use and its related emissions by households are determined by the budget categories electricity, natural gas, and so on. The emissions of other consumer items are only taken into account in the basic goods and not in the use phase as such.4 As a result, all environmental loads are added (1,283 CO2 equivalents [CO2-eq]),5 and the intensities are determined by dividing the total by the price (0.824 GWP/€) and by dividing it by the weight (1.283 GWP/kilogram [kg]). All environmental intensities (load/€) for 368 household expenditures have been calculated this way. These 368 household expenditures match those of the Dutch national consumer expenditure survey (CBS 2002). By combining the environmental intensities with these expenditure data, the environmental load of an expenditure and the total environmental load for an average Dutch household can be calculated.

For more background articles regarding a description of the methodology, please refer to Engelenburg and colleagues (1994), Wilting (1996), and Kok and colleagues (2006).

To use the hybrid method described above, many data are required. These data can be divided into physical and economic data, both linked to environmental data. The physical data contain databases for the basic goods, for example, a list of 169 goods with environmental data per kilogram, packaging materials, transport modes and waste processing. These data are obtained from several life cycle assessment (LCA) studies and databases such as the European Life Cycle Data (ELCD) database (European Commission 2007; Blonk et al. 2008); the Eidgenössische Technische Hochschule in Zürich (ETH3) database; United Nations Food and Agriculture Organization (FAO) statistics (land use); the Livestock, Environment and Development Initiative (LEAD; livestock), IDEMAT, and Bundesamt für Umwelt, Wald und Landschaft (Swiss Agency for the Environment, Forests and Landscape) (BUWAL).

The economic data contain IO tables and trade statistics (margins) as well as manufacturer and sector statistics (value added, depreciation, energy consumption, production). Environmental data were obtained for all sectors from the database of the Dynamic Input-output Model to study the Impacts of Technology Related Innovations (DIMITRI) model, which obtains data from, inter alia, energy statistics and the pollutant release and transfer register (Idenburg and Wilting 2004).

For the GWP impact category, three substances are included: CO2, CH4, and N2O, with their specific conversion factors (see table 2). These greenhouse gas emissions data are obtained from the DIMITRI model into which mobile, stationary, and process emissions are incorporated. An important source of nonenergy CO2 emissions is the production of building materials. Important sources of CH4 are animal husbandry (food digestion and manure) and landfills. Important sources of N2O are nylon and nitric acid production and land-use-based emissions of agriculture due to fertilization.

Table 2. Conversion factors from stressors to impact category equivalents per unit of weight
Impact categories CO2 CH4 N2O N P SOx NOx NH3 VOC Land use
  1. Source: Guinée (2002).

  2. Note: CO2= carbon dioxide; CH4= methane; N2O = nitrous oxide; N = nitrogen; P = phosphorus; SOx= sulfur oxides; NOx= nitrogen oxides; NH3= ammonia; VOC = volatile organic compound; GWP = global warming potential; CO2-eq = carbon dioxide equivalents; SO2-eq = sulfur dioxide equivalents; PO43−-eq = phosphate equivalents; VOC-eq = volatile organic compound equivalents; ha = hectares.

GWP (CO2-eq)125296       
Acidification (SO2-eq)     1.000.701.88  
Eutrophication (PO43−-eq)   0.423.06 0.130.35  
Summer smog (VOC-eq)      1.93 1.00 
Land use (ha)         1.00

For the impact category acidification, the emissions of SOx and NOx are mainly related to energy use, and the emission of NH3 is determined by practices in animal husbandry and agriculture (especially the handling of manure). The impact category summer smog is also partly energy related: NOx (98%) and NMVOC (55%) emissions originate largely from combustion processes. For the impact category eutrophication, emissions of nitrogen and phosphate (mainly from agriculture), and to a lesser extent NH3 and NOx, are important.


In this section, all results shown are Dutch household averages. The energy, land use, and environmental impact categories have been calculated for 368 consumer expenditures, as described in the previous section. The stressors used are energy, land use, and the emissions of CH4, N2O, SOx, NOx, NH3, volatile organic compounds (VOCs), aerosols (PM10), nitrates, and phosphates. These stressors are used to calculate the GWP (i.e., total emission of greenhouse gases in CO2-eq), acidification (in SO2 equivalents [SO2-eq]), eutrophication (in phosphate equivalents [PO43−-eq]), and the contribution to summer smog (photochemical oxidation in VOC equivalents [VOC-eq]) according to the equivalence factors from Guinée (2002).

We combined the 368 consumer expenditure items into 12 domains using the COICOP classification (UN 2010):

  • 1Food: Food and nonalcoholic beverages. Fresh and precooked food purchased for consumption at home.
  • 2Alcohol + tobacco: Alcoholic beverages and tobacco purchased for consumption at home.
  • 3Clothing: Clothing, footwear, and accessories, including cleaning, repair, and hire.
  • 4Housing: Housing, water, electricity, gas, and other fuels.
  • 5Furnishings: Furnishings, household equipment, and routine maintenance of the house.
  • 6Health: Health products, appliances, equipment, and services.
  • 7Transport: Purchase, use, and maintenance of vehicles and public transport.
  • 8Communication: Postal, telephone, and telefax equipment and services.
  • 9Recreation: Recreation and culture.
  • 10Education: Preprimary to tertiary education for young people and adults.
  • 11Restaurants: Restaurants, hotels, and other accommodation services.
  • 12Others: Miscellaneous goods and services, including personal care, jewelry, insurance, and other services.

Figure 2 presents the expenditures of Dutch households. Four domains (food [excluding preparation], housing, transport, and recreation) cover about two-thirds of total Dutch expenditures.

Figure 2.

Household expenditures, the Netherlands, 2000; tob = tobacco.

Figure 3a presents the energy consumption of the 12 domains. Housing (41%), transport (17%), recreation (14%), and food (11%) have the largest share, which is similar to expenditures. The same applies, more or less, for GWP. However, there is a clear and meaningful difference: the GWP of food here is in second place (19%) (see figure 3b).

Figure 3.

Energy consumption and the five impact categories for 12 expenditure domains in an average Dutch household. The 12 domains are derived from the United Nations (UN) reference, Classification of Individual Consumption according to Purpose (COICOP).

The same four domains have the largest contribution in all impact categories except eutrophication. The order of these four domains is different, but they are responsible for about 85% of the total impact. Food plays an important role in all five impact categories. Food makes the largest contribution in the impact categories acidification (41%), land use (62%), and eutrophication (84%) (see figures 3c, f, and d, respectively).

The main conclusion from the results shown in figure 3 is that food has by far the largest total impact of all consumer domains, while it ranks fourth in expenditures. This conclusion is confirmed by the results presented in figure 4, where all of the percentages from figure 3b to f are added, unweighted. By doing this, we do not mean for the five impact categories to be equally important. Figure 4 is just an illustration to show, in one picture, that food scores high or highest in all five impact categories, and only if GWP and/or summer smog are weighted much more heavily will another domain become dominant.

Figure 4.

Cumulative percentage contribution of all environmental impact categories in all consumer domains.

The conclusion that can be drawn from the results above is that overall the food domain has the largest environmental impact. The environmental impact of food is present at all levels: local or national (eutrophication, summer smog), regional (acidification), and European and global (GWP and land use). The importance of food has also been shown in other studies by Carlsson-Kanyama and colleagues (2003), Nijdam and colleagues (2005), Tukker and colleagues (2005), and Risku-Norja and Mäenpää (2007).

Food is a heterogeneous group; it consists of 111 expenditures varying from whole wheat bread to wine, from carrots to beef, and from bananas to restaurant meals. We grouped the 111 expenditures into 11 COICOP subdomains (UN 2010): bread and cereals; meat; fish and seafood; milk, cheese, and eggs; oils and fats; fruit; vegetables; sugar, jam, honey, chocolate, and confectionery; food products not elsewhere classified (n.e.c.); coffee, tea, and cocoa; mineral waters, soft drinks, fruit, and vegetable juices.

Figure 5 presents expenditures in the 11 COICOP food subdomains. Bread, meat, dairy, and vegetables are the four largest subdomains, which together consume about 68% of the total food budget. Although in a different order, the same applies to the energy graph (figure 6a), which greatly resembles the expenditure graph (figure 5).

Figure 5.

Expenditure within the food domain.

Figure 6.

Energy consumption and the five impact categories for the 11 food subdomain expenditures in an average Dutch household.

The four largest subdomains of the expenditure graph also make a large contribution to most of the five impact graphs, about 69% combined on average. Fats have a large share in the eutrophication impact category, and the summer smog impact category deviates the most from this pattern, with the largest share coming from dairy, fruit, soft drinks, and bread. Figure 6d shows an almost 40 % share of eutrophication from the dairy subdomain, which is more than three times higher than the meat subdomain.

Figure 7 presents an overview graph, similar to figure 4, that adds up the share of the contributions to the five impact categories for Dutch food expenditures.

Figure 7.

Cumulative percentage contribution of all environmental impact categories in the food domains.

So far we have only looked at expenditure domains. With the applied method, it is possible to look at an even higher level of detail: the product level. This study did this for 368 expenditures. Table 3 and table 4 each present four top 10 lists. Table 3 presents ten itemized consumer expenditures (food and overall) that contribute the highest total energy loads (columns 1 and 2) and that contribute the highest cumulative environmental impacts across all five categories (columns 3 and 4). Table 4 presents the same information measured not by cumulative contribution but by intensity (load or impact per euro).6

Table 3. Four lists on a product and product group level showing the top ten contributors of energy and environmental loads
  Energy, food expenditures Energy, all expenditures All impacts, food expenditures All impacts, all expenditures
 1CheeseNatural gasCheeseGasoline, oil, cars, motorcycles
 2Dine outGasoline, oil, cars, motorcyclesMilkCheese
 3Dine/drink out not specifiedElectricityPotatoesNatural gas
 4Other nonalcoholic beveragesOther vacation costs abroadMinced meat, freshElectricity
 5Fish, freshOrganized vacation trips abroadCake, cookiesRental value
 6MilkRental valuePastryRent
 7PastryCarOther sausages/meat productsMilk
 8Cake, cookiesRentPoultryOther vacation costs abroad
 9WineCollective energy costs, central heatingFruit juiceOrganized vacation trips abroad
10Whole wheat breadCheeseWhole wheat breadPotatoes
Table 4. Four top ten lists for energy and environmental intensitya of a product or product group level
  Energy, food expenditures Energy, all expenditures All impacts, food expenditures All impacts, all expenditures
  1. Notes: aThe environmental intensity as calculated in this table is the sum of the relative intensity (intensity/average intensity) for each impact category, all based on monetary expenditures.

  2. bMainly fuel wood, used in fireplaces, with extremely high volatile organic compound (VOC) emissions.

 1Fish, freshNatural gasSalad oilSolid and liquid fuelsb
 2Salad oilCollective energy costs, central heatingFats for fryingSalad oil
 3Other fishSolid and liquid fuelsbButterFats for frying
 4Herring, saltedFish, freshEggsButter
 5Frozen vegetablesElectricityPotato starchEggs
 6Dried vegetablesEnergy costs included in rentPotatoesPotato starch
 7TomatoesGasoline, oil for cars and motorcyclesCheesePotatoes
 8Preserved fishOther gasoline and oilCreamCheese
 9Fats for fryingSalad oilOther dairy products not specifiedCream
10Fruit, compote/pureeOther fishFish, freshOther dairy products not specified

The rankings in table 3 are dominated by products with a high energy or environmental intensity or high expenditures (and sometimes both). Impacts across all products (columns 2 and 4) are dominated by large expenditure items such as direct energy (natural gas, gasoline, electricity), vacations, and housing (rental value, rent). Mixed into these lists are a few food products with high environmental loads. While the number of food items is about 30% of the 368 items considered, and while food expenditures account for only 12% of total expenditures, there are still 3 food items present in table 3 (column 4). Dine out and dine/drink out not specified are present in column 1 but not column 3, because these items have especially high energy inputs (space heating and/or cooking), while other environmental impacts are lower.

In table 4, the changes in ranking between energy and all impacts (columns 2 and 4) speak for themselves. Almost all items in the all impacts column (4) are from the food domain; this is in contrast with the top ten energy impact products in column 2, where direct energy items are strongly represented. The exception in column 4 is solid and liquid fuels. This expenditure category contains mainly fuel wood used in fireplaces that have extremely high VOC emissions and therefore rank highest in overall environmental intensity.


The hybrid energy analysis method could easily be extended to become a broader environmental analysis method. Analyses can be made for several cross sections of household consumption patterns, even at the level of single household expenditure categories (see table 3, table 4, and the appendix available as supporting information on the journal's Web site). This is what distinguishes the hybrid method from the IO method (see Kok et al. 2006).

Although energy can be used as a proxy for the environmental pressures from household consumption, other indicators provide new insights. Among the 12 defined expenditure groups in this article, food is the most sensitive to additional impact categories. The share of food in energy use is 11%, but is substantially higher in the impact categories acidification (41%), land use (62%), and eutrophication (84%). For acidification, NH3 emissions from livestock is are explanatory factor. The share of food in the impact category GWP is two-thirds higher than in energy use due to N2O and CH4 emissions.

The consumption of large amounts of dairy products like milk and cheese explains the results described below.

Within the food domain, livestock-related (meat, eggs, and dairy) consumption causes a substantial part of the environmental load. This is especially the case in the impact categories land use, eutrophication and acidification, with shares in the food domain of 43%, 51%, and 60%, respectively. Beef and milk (bovine animals) contribute a large share of land use impacts, which is partly explained by the lower dressing factor, or fraction of the animal that can be used for human consumption (Elferink and Nonhebel 2007). The large contribution by dairy to eutrophication (see figure 6d) can be explained by the use of fertilizer on grassland for dairy cattle in the Netherlands. A substantial portion of meat comes from other animals (pigs, chickens), which are fed with fodder from abroad for which, generally, less or no fertilizer is used. NH3 makes the largest contribution to the acidification impact category. Pork is a large contributor here, but cheese is by far the largest. The NH3 emissions from cheese are responsible for 11% of total acidification. Weber and Matthews (2008a) found red meat in the United States to be a major contributor to GWP, with dairy products in second place (among food products exclusively). More meat in the United States and more dairy products in the Netherlands, products of eating habits, will probably explain this difference.

Oil and fats have a high energy and environmental intensity when looking at expenditure levels (table 4), while cheese is ranked high when looking at energy and environmental load (table 3). Both are present in all four rankings. Other items with high intensities are direct energy (natural gas, gasoline, and electricity), potatoes, fish, and dairy products generally. Items with a high load include direct energy (natural gas, gasoline, and electricity), vacations, and housing.


The hybrid methodology has been evaluated and compared with other methods exclusively on energy use (Kok et al. 2006), with the conclusion that all methods have their specific advantages and disadvantages. The strength of the hybrid method is better results on a product level. The methods and data sets have proven accurate (see Wilting 1996). Wilting calculated an uncertainty of 7.7% due to the energy data used and another 10% due to uncertainties in prices. However, emissions statistics are not as accurate as energy statistics, so additional uncertainty is introduced here. To give an idea, Olivier and colleagues (2009) determined that energy-based sector CO2 emissions have an uncertainty of ±3%, compared with ±5% for GWP emissions. Gijlswijk and colleagues (2004) did similar calculations for acidification and calculated an uncertainty of ±10%. The other impact categories are expected to have similar uncertainties because they partially use the same emissions data. When the results are used in a mutual comparison of household expenditures, uncertainties are less important.

The hybrid method has two major disadvantages. The first disadvantage is the truncation of the LCA data used; this may cause incomplete results for basic goods (Suh et al. 2004). The second disadvantage is the assumption that imports are produced as they are domestically, at least for the IO-derived data (Engelenburg et al. 1994). Although this assumption has been a very common one, an increasing number of studies try to incorporate imports into the methodology. (See Tukker and Jansen [2006] for an overview.) The applied method takes imports partly into consideration. The energy and emissions of the basic goods are based on international data, and the environmental transport costs from the manufacturer to the trade sector are taken into account (based on international import statistics). An exception is made for vegetables and flowers, which, in the Netherlands, are produced mainly in the horticultural sector. Imported vegetables and flowers are assumed to come from open-air horticulture instead of coming partly from heated greenhouses. What remains is the production itself. Electricity use, in particular, and how this electricity is produced, will influence the final results: it will make a big difference if a product is produced in Norway (mainly hydroelectricity, assuming no conversion losses and no CO2 emissions) or in Greece (where energy production is mainly fossil fuel based [EurElectric 2000]).

Vringer and colleagues (2010) compared the hybrid method applied here with a single-region (SRIO) and a multiregion (MRIO) IO method. Results from this comparison show that the impact values of the hybrid method are in between the SRIO and the MRIO.

The environmental stressors used determine at least in part the outcome that the food category has the highest environmental impact. If other environmental stressors are used, such as carcinogenic to humans and toxic substances (see Morris and Matthews 2010), other outcomes can be expected.


We would like to thank Harry Wilting, Kees Vringer, Eric Drissen, Nico Hoogervorst, and Corjan Brink for their cooperation in establishing a consistent and validated data set. Furthermore, we would like to thank Rixt Kok for her assistance with the EAP methodology. The authors would like to acknowledge the anonymous reviewers for their valuable comments.


  • 1

    Classification of Individual Consumption according to Purpose (COICOP) is a reference classification published by the United Nations (UN) that divides the purpose of individual consumption expenditures.

  • 2

    For more background information about the IO process and hybrid analysis please refer to Suh and colleagues (2004), Suh and Kagawa (2005), Hendrickson and colleagues (1998), Kok and colleagues (2006), and Hondo and colleagues (2002).

  • 3

    The hybrid method was implemented in the energy analysis software program (EAP) developed by Wilting, Kok, and Benders (Wilting 1996; Wilting et al. 1999; Benders et al. 2001; Kok et al. 2001).

  • 4

    For example, the analysis of budget item “wallpaper and painting costs” contains the basic goods acrylic and alkyd resin paint. The emission values used for these paints incorporate the VOC emissions in the use phase.

  • 5

    Carbon dioxide equivalent (CO2-eq) is a measure for describing the climate-forcing strength of a quantity of greenhouse gases using the functionally equivalent amount of carbon dioxide as the reference.

  • 6

    The environmental load and/or intensity is the total of all impact category shares per consumer item. For each expenditure item the relative contribution is calculated by dividing the intensity by the average of all intensities. For each expenditure item, this relative contribution for the five impact categories (GWP, acidification, eutrophication, summer smog, and land use) is added unweighted.

About the Authors

René M. J. Benders is a researcher at the Center for Energy and Environmental Studies at the University of Groningen. Henri C. Moll is a professor of resource management for sustainable production and consumption at the Center for Energy and Environmental Studies at the University of Groningen. Durk S. Nijdam is a researcher at the Environmental Assessment Agency (PBL) in Bilthoven, the Netherlands.