Consumption-based emissions for municipal planning

Cities contribute widely to the global greenhouse gas (GHG) emissions. In Europe, 74 % of inhabitants live in cities, and urban settlements contribute between 60-80 % to the global GHG emissions (Reckien et al., 2018). Cities can also be important actors in reducing the emissions, and a large number of cities globally have made commitments to reduce GHG emissions. Indeed, in Europe, 80 % of cities with more than 500,000 habitants have a climate change mitigation or adaptation plan (Reckien et al. 2018). In order to effectively set targets and monitor progress towards them, measuring city-level emissions regularly is essential.

Two principal approaches exist to calculate local-level emissions: production-based or territorial approach, and consumption-based approach. The former refers to the emissions generated by the production and activities within the territorial area in concern, such as energy generation, traffic or emissions from waste. An alternative approach would be to study consumption-based emissions, that are emissions from the products and services consumed in an area taking into account the upward emissions from the supply chain of these products and services independently of their origin.

Traditionally, city or local level emissions have been measured through the production-based approach. For one, obtaining information of the emissions produced within the geographical area in question is easier than estimating emissions from the whole supply chain. In addition, it has previously been thought that cities or municipalities would have only limited influence over emissions produced outside their territory but caused by consumption of its inhabitants. However, recent climate strategies and roadmaps increasingly acknowledge the possibilities of local authorities to influence the consumption-based emissions directly or indirectly within their jurisdiction.

Local level emissions may differ widely based on the approach utilised for measuring the emissions. A report compiled by the C40 Cities Climate Leadership Group (C40) about consumption-based emissions in 79 cities concluded that consumption-based approach revealed on average 60 % higher emissions attributable to the cities compared to the production-based approach. However, the cities can be divided into two categories, “producer-cities” which produce large quantities of products for exports and whose production-based emissions surpass consumption-based emissions, and “consumer-cities” that are dependent on imports and whose consumption-based emissions are greater than the production-based emissions. Most of the producer cities in this study were in Southeast Asia and Africa, while the consumer cities were mostly found in Europe and North America (C40 et al., 2018).

Thus, consumption-based emissions comprise a significant amount of the GHG emissions in cities in more developed countries and leaving them out from local greenhouse gas inventories might provide a one-sided understanding about local level emissions or possibilities to reduce them. Many academic studies highlight the need to introduce consumption-based carbon accounting alongside measuring territorial emissions to better understand the role and impact of local and national level policies to emissions reduction (see Ottelin et al., 2019, Ottelin et al., 2018a). Consumption-based emissions accounting may, for instance, reveal a so called carbon leakage, which is referred to relocation of emitting activity from one area to another, that will simply lead to a decrease emissions in one area while increasing them in another one.

Local level emissions are often categorised in three “scopes”. Scope 1 refers to GHG emissions from sources that are located within city boundaries, Scope 2 to consumption of grid-supplied energy sourced from outside the city, and Scope 3 refers to emissions outside the city boundaries as a result of activities taking place within the city, including waste and waste water management, energy transmission and emissions from consumption (WRI et al., 2014). Based on this division, consumption-based emissions are sometimes referred to as Scope 3 emissions in different contexts, even if in many traditional production-based emissions calculations, only Scope 3 emissions related to waste, energy and transport are included.

3.1. Main approaches for measuring consumption-based emissions at a local level

Overall, consumption-based accounting refers to allocation of all the production and delivery chain emissions to the end-user (Heinonen et al., 2020), that is to say emissions are allocated based on actors’ use of goods and services (Balouktsi, 2020), who in the case of a local level accounting are the consumers within a city or municipality.

Consumption-based emissions can be measured from two different perspectives: from top-down or bottom-up approaches, and in addition through hybrid approaches that have characteristics from both approaches. Each have their advantages and disadvantages, and a suitable method should be selected based on purpose and use of the information compiled. In this section, the principal approaches to local level emission calculations are presented.

Top-down approaches (Heinonen et al., 2020) or estimations based on spending data (Broekhoff et al., 2019), utilise input-output (IO) models that are tables utilised for economic analysis, such as national accounting, and they include information on production, imports and exports for different product groups and sectors. Environmentally extended input-output models provide emissions intensities per amount spent for a particular product group. Based on the tables, emissions produced per dollar spent per consumption category can be estimated (Broekhoff et al., 2019). These models increasingly provide information at a sub-national level, and according to Heinonen et al. (2020) the accuracy of the models have increased in recent years. Often, however, an input-output table or database has information available only at a national level, in which case consumption surveys can be applied to scale down the information into local level.

The basic model for accounting for consumption-based emissions based on IO models is depicted in Figure 1.

Some of the disadvantages of this model include the complexity associated with utilising vast input-output tables. Moreover, the information often needs to be adjusted to a local level which may lead to difficulties in reconciling local level data with high-level IO tables leading to robust estimations (Balouktsi, 2020). The models estimate an average emissions intensity for a category of consumption, which may neither be sensitive to developments in an emissions intensity of an industry or service, nor is the model sensitive to consumer preferences to less carbon intensive brands over more carbon-intensive ones. In addition, the data available is often dated back even for years and may not thus represent the latest developments in emissions efficiency of an industry (Heinonen et al., 2020).

Bottom-up approaches include local consumption surveys and other sources of information such as waste audits, utility billing data and so on, combined with life-cycle analyses (LCA) to estimate emissions based on physical units of consumption. Depending on the availability of information, estimating this may be either a daunting task or a reasonable exercise: having information on the emissions intensity and level of consumption are crucial in being able to utilise this approach. Balouktsi (2020) and Broekhoff et al. (2019) both basically discard the feasibility of a full process-based LCA approach at a city scale. Sometimes national household emissions patterns are scaled down to the local level, but these estimations are rough and may omit specific local patterns of consumption.

In addition to mere top-down or bottom-up approaches, hybrid models can be utilised. Basically, hybrid models utilise data from input-output analyses to get an overall view on emissions and include more detailed life-cycle analysis data for a part of the products or services. When conducting a hybrid analysis, specific attention needs to be paid to the comparability of data. Different datasets may have differing underpinning assumptions that make comparison difficult; for example, some input-output tables may utilise destination-based accounting while consumption data may be residence-based (see more about methodological choices in chapter 3.2). However, hybrid models may be functional and increase the accuracy of results or sensitivity of data about change (Broekhoff et al., 2019) when constructed carefully, being transparent about methodological choices made.

3.2. General underpinnings of local level consumption-based emission accounting

Consumption-based emissions inventories are essentially models, rough estimations or approximations about the emissions related to consumption in a specific area. When estimating consumption-based emissions, there are several methodological choices to be made that will have an impact on the results. Heinonen et al. (2020) reveal that methodological choices may lead to a difference of up to 80 % in the estimated emissions. Thus, being explicit with the choices made is paramount.

Emissions accounting may be done from two different perspectives: residence-based accounting or destination-based accounting. Residence-based accounting allocates “to a consumer the GHG emissions caused by his/ her consumption regardless of the geographic location of the occurrence of the emissions” (Heinonen et al., 2020). Destination-based accounting, on the other hand, allocates to the studied location the emissions from final purchases on the location, including the emissions from global supply chain and production, independently of the original residence of the persons (Heinonen et al., 2020). In the latter case, local emissions may be distorted for example by a high level of tourism, or if a commuting city receives high numbers of daily commuters.

In their analysis of 111 academic articles on consumption-based emissions estimations Heinonen et al. (2020) conclude that main factors contributing to differing results for consumption-based emission inventories include the approach selected (residence- or destination-based), and scope of the study, especially related to whether governmental and private sector consumption is included. Heinonen et al. (2020) distinguish between “personal carbon footprint” and “areal carbon footprint” based on whose consumption is accounted for: the former includes only the consumption of the residents of the city, while the latter includes additionally government consumption and private investment. Including or excluding the latter may influence the results by 10-50 %, thus omitting the two other consumer groups may distort the results considerably.

In addition, methodological choices such as the number of GHGs accounted for, inclusion or exclusion of land-use change, as well as methods for accounting for direct energy use (for example heating costs may be hidden in housing bills), and ways in which emissions from durable goods are allocated to the end-user all influence the results. Finally, imports are included and accounted for very differently in many of the studies. Often, imports are given the same emissions coefficients as national production, which may be misleading (Heinonen et al. 2020).

Balouktsi (2020) brings forth also the categorisation of emissions into differing groupings in different studies. Lack of standards in reporting reduces comparability of information as well as verifiability, which would both be important for a municipality.

Further differences are caused for example by the presentation of the data: it can be presented only as total consumption-based emissions, or emissions per household, per capita or per consumption unit. Some studies differentiate between domestic, public and private sector consumption, while others present them all blended together.

Since the field is relatively new, especially at the municipal level, there is still a big variation in how the emissions inventories are conducted, which makes it difficult for a city or municipality interested in the topic to conduct the process. Few guidelines or protocols seek to respond to this need, and they are presented further down in this chapter.

3.3. Existing standards and methodologies

While quite a few standards and methodologies exist for accounting production-based emissions at municipal level, only a few of them take into account indirect emissions from consumption – or Scope 3 emissions. In this chapter, main existing standards and guidelines produced by different organisations or research institutions are presented. Each of the selected standards and guidelines have been applied in practice in municipal level consumption-based emissions analysis.

In addition, a range of academic research analysing different methods and approaches to consumption-based emissions accounting is available. It was, however, out of the scope of this study to analyse and present them here.

3.3.1. British Standards Institution - PAS 2070

British Standards Institution has developed a standard for assessing carbon emissions at a city or municipal level. The standard covers both direct and indirect GHG emissions. The aim is to provide “consistent, comparable and relevant quantification, attribution and reporting of city-scale GHG emissions” (BSI, 2013).

The consumption-based (CB) methodology covers direct and life-cycle GHG emissions for all goods and services consumed by residents of a city. The approach is residence-based, and thus does not consider emissions generated by visitor activities or services provided to visitors. The methodology utilises a top-down approach in that it recommends the utilisation of environmentally extended input-output model based on financial flow data from national or regional economic accounts, combined with environmental account data.

The data required will include household consumption, municipal and national government (for residents benefitting from government expenditure within the city boundary), and business capital expenditure on goods and services within the city boundary.

A full emissions inventory was conducted in London as a practical application of the standard, including territorial emissions and consumption-based emissions for six greenhouse gases. The first strategy reported per capita emissions of 10.05 t CO2e/ per capita while the consumption-based approach yielded up to 14.15 t CO2e/ per capita. Total consumption-based GHG emissions in London were, thus, 40 % higher than the production-based emissions (BSI, 2014).

The study presents the results for household consumption in six quite general categories, including food and drink, utility services, household, transport services, private services and other goods and services – a typical categorisation for a consumption-based emissions inventory. However, such broad categories will give quite a limited understanding about any possible change should the study be repeated in the future. The document also presents a second analysis based on COICOP categorisation often utilised in consumption-based emissions analyses, based on 12 main categories and 40 sub-categories (COICOP is the classification of the United Nations Statistics Division for individual consumption according to purpose). However, even with this categorisation, the main emissions come from “transport services” and “electricity, gas and other fuels” and “food”. More detailed information from these categories might provide useful insight for identifying “hotspot” areas for policy planning, if the information was provided.

The report disaggregates the results also by CO2 and non-CO2 gases. The division is not common in consumption-based emissions accounting but might be interesting for policy purposes, given that for example for food, most greenhouse gas emissions come from non-CO2 emissions.

3.3.2. ICLEI USA Community-based emissions standard

Local Governments for Sustainability (ICLEI) is an international network of local governments pursuing sustainability and taking action to reduce their carbon footprint. ICLEI USA is an independent regional network of the broader international ICLEI community.

In 2013, ICLEI USA produced a protocol for local governments to account for their carbon footprints; “the U.S. Community Protocol for Accounting and Reporting of Greenhouse Gas Emissions”. The protocol is based on an international emissions standard created by World Resources Institute, C40 and ICLEI “the Global Protocol for Community-Scale Emissions (GPC)” (WRI et al. 2014) but is especially directed to the US cities. The protocol includes a voluntary section for accounting consumption-based emissions. ICLEI USA also provides training for communities on how to start their consumption-based emissions inventories (ICLEI USA, 2020).

Appendix I of the protocol presents three different alternatives for conducting a consumption-based emissions inventory at a local level. Overall, the protocol takes a residence-based approach to emissions calculation, recommending to account for consumption patterns of residents of a community, independently of the place where the emissions occur. The basic principle behind emissions accounting is the following (Fig. 2):

As a first approach, the protocol recommends the possibility to utilise the CoolClimate Calculator of University of California, Berkeley (2020) that provides information of an average carbon footprint at a community level for a wide range of communities in the US. This information can be then multiplied by the number of inhabitants. The data for the calculator corresponds to the year 2008, and various data sources have been utilised to compile it. Needless to say, the approach provides a rough estimation of household consumption only and is thus recommended mainly for small municipalities with limited resources. If a community chooses to utilise this approach, the protocol recommends running the information against community’s own data sets, regarding for example emissions intensity of energy utilised, to increase accuracy (ICLEI USA, 2019).

Household surveys are another option to approach consumption-based emissions. This approach would require a municipality to either conduct a household consumption survey, or in other form have the required information at hand, such as through a national consumption survey, and then apply emissions factors to the amounts of each purchase. This is presented as a costly exercise requiring a significant level of expertise (ICLEI USA, 2019).

The third approach is to utilise a customised model utilising macroeconomic data. IMPLAN software is provided as an example – IMPLAN is essentially an input-output model covering all private industries in the US, including inter-state and international trade classified by the North American Industry Classification System (NAICS) codes (IMPLAN, 2020). While this is also a costly exercise requiring specific expertise, it allows municipalities to obtain more accurate data with detailed resolution for types of consumption. The approach includes consumption of the governmental entities and business capital investment, providing, thus, a “full” consumption-based emissions inventory (ICLEI USA, 2019).

3.3.3. Stockholm Environment Institute guidelines

Stockholm Environment Institute (SEI) has engaged in various consumption-based emissions inventory processes, and it has published guidelines or recommendations for cities or municipalities that are interested in conducting such processes. The recommendations, presented in Broekhoff et al. (2019), endorse the use of input-output tables either by spending or by physical unit data, and utilisation of consumer spending data, if available, to scale national-level information down to a local level to create a reasonably accurate emissions inventory. National level consumption data can be downscaled to a local level, but this will affect accuracy of the report.

The report also analyses the use of life-cycle analysis tables, deeming the exercise as complicated, unless relevant data is readily available (Broekhoff et al.,2019).

Moreover, the document briefly introduces simpler but less accurate ways to estimate consumption-based emission footprints. These include analysis of waste data, which is often responsive to behaviour change in consumption but leaves out many sectors of consumption – including services and transport. Broekhoff et al. (2019) also mention the possibility to base the analysis on data available from CoolClimate data or similar calculation tools – if available for the region in question. In simplest form, a city can also use existing national household consumption emissions studies and expect the same patterns of consumption to hold roughly at a city level (Broekhoff et al., 2019).

Overall, Broekhoff et al. (2019) put an emphasis on the purpose and intended use of the inventory as factors that should guide the selection of the methodology and methodological choices made. Different data sources can be utilised depending on whether the main aim is to engage in policy follow up, citizen engagement or overall data comparison between municipalities. For example, while hybrid approaches utilising a series of local level data may gain in accuracy, they will lose in comparability with other cities.

The recommendations are given at a general level. Thus, no specifications on such details as which greenhouse gases to include, or the scope of the inventory, are defined. No stance is either provided on the selection of a general approach – residence or area-based, or whether to include governmental consumption and private investment into local consumption-based emission inventories.

3.4. Possible data sources for a municipal consumption-based emissions inventory

Availability of data influences and directs the possible methodology to be utilised, alongside other considerations, such as the final use of the information.

The use of input-output tables is recommended by most of the standards and guidelines while input-output tables by spending unit (as opposite to physical unit) are most widely used for the emissions inventories (for example Heinonen et al. 2020). Each of the tables or databases have a slightly different methodological approach. For instance, their geographical focus might vary, or the scope and data resolution may be different. Some of the databases require a licence, and the prices may vary.

As a rule of thumb, one could argue that international multi-region input-output tables are especially suitable when international comparability of results will be important, while single-region tables, such as the Finnish ENVIMAT (Table 1), may provide more accurate results within the region it is designed for.

None of these input-output tables provide information scaled down to a municipal level. Hence, information about consumption at a local level needs to be obtained from other sources, to be then combined with the relevant input-output data. Basically, both academic and non-academic studies on consumption-based emissions in Finland have so far relied on data from Statistics Finland's consumption surveys that are repeated and released every four years. For example, Finnish Environment Institute (SYKE) utilises consumption data from 2016 in the most recent survey of consumption-based emissions in Finland (Nissinen and Savolainen, 2019). In the absence of consumption data, municipalities have conducted their own consumption surveys, as it will be discussed further along in this report.

While input-output tables and consumption data are the main components of a consumption-based emissions study based on top-down approach, other municipal level data will be required. For example, C40 et al. (2018) illustrate data sources utilised for their study, which include additional data on private transport and household energy emissions data, as well as data on city industrial output to scale down national emissions factors for industrial production.

Both Broekhoff et al. (2019) and Balouktsi (2020) raise a concern that a municipality might easily focus too much time and resources on conducting a high-accuracy consumption-based emissions inventory instead of maintaining the focus on actual implementation of activities that focus on reducing emissions from consumption. Thus, they suggest that finding more simple solutions to measure such emissions can be a viable option. These may include studying waste data or different carbon footprint databases.

In the case of Finland, supermarket chains have developed their own carbon footprint calculators available for the members of their loyalty programmes. Members of Kesko’s loyalty programme can now follow up the carbon footprint of the items purchased at a product group level. The footprint calculator is based on Natural Resources Institute Finland’s (LUKE) LCA analyses for different product categories (Kesko, 2019). Similarly, another big Finnish grocery store chain, S-ryhmä, published its own calculator for the members of their loyalty programme, similarly utilising the LCA data from LUKE (S-ryhmä, 2019). Based on the statistics of Finnish Grocery Trade Association (PTY 2019), the combined market share of these two supermarket chains was 83 % in 2019 in Finland. If it was possible to combine the consumption data of these two actors on a local level, estimations of consumption and emissions might be possible to draw at least for certain consumption categories.

Another approach worth assessing would be to analyse the data from a specific carbon footprint calculator directed to consumers to determine typical consumption patterns, and to extend the results to the whole population of the city. Some considerations would be necessary; for instance, the persons that tend to respond to such calculators, tend to already be environmentally more conscious than an average consumer, and the results would need to be adjusted to be more representative of the population of the area.