Calibration
The ThreeME database is built from a number of databases, mainly from Eurostat. The resulting characteristics from the ThreeME database, and thus the ThreeME model, are summarized below. These characteristics refer to the maximum level of detail possible in ThreeME. The actual level of detail as applied in case studies can be lower as for each simulation in ThreeME, the database is aggregated to meet the needs of the research question at hand.
| Dimension | Coverage |
|---|---|
| Year | 2015 |
| Regions | EU27_2020, FRA and NLD |
| Currency | EUR |
| Sectors | 97 sectors (65 NACE Rev 2 sectors + 28 energy sectors + 4 transport sectors) |
| Products | 89 products (65 CPA2008 products + 20 energy products + 4 transport products) |
| Energy in Mtoe | Supply and use by 20 products and 97 sectors, plus household use, imports and exports |
| GHG Emissions | 6 emissions (CO2, CH4, N2O, NF3, SF6, HFC and PFC), 20 energy products, 97 sectors and 3 source secors (conbustion, production and material) |
The data calibration is done using aflexible tool developed in the R programming language. This tool has several advantages. First, all assumptions are fully traceable. It replicates the ThreeME database using only the raw data files in their original format as directly downloaded from their data source. Thus, all data processing assumptions can be found in the R code. Second, the tool has a flexible setup with built-in options for adjusting the data source, year, country, sectors and products. As a result the database can be customized for each ThreeME simulation in a relatively easy manner.
The databases on which ThreeME is built are listed in the table below. The calibration of each Computable General Equilibrium (CGE) model starts with the Supply and Use Table (SUT) as this contains the economic transactions between agents such as governments, sectors and households. The ThreeME SUT is combined with non-financial transactions data to capture the redistribution of incomes between households and government. Then, Energy Balances are use to create the full layer of energy supply and use in Mtoe, specified by product as well as sector. The same energy data in Mtoe is converted to euro in order to increase the number of energy products and sectors in the SUT. For the conversion IEA energy prices are applied. GHG emissions in Mton are added in the next step. Also, the number of transport sectors are increased in order to distinguish rail from road transport. This is done using Eurostat Structural Business Statistics. Finally, the investments matrix is estimated from Eurostat data. Each of these steps are described in more detail in the following sections.
| Data | Source(s) |
|---|---|
| Supply and use tables | Eurostat supply and use tables: naio_10_cp15, naio_10_cp16, naio_10_cp1610, naio_10_cp1620, naio_10_cp1630 |
| Redistribution of incomes | Eurostat non financial transactions: nasa_10_nf_tr |
| Energy in Mtoe | Eurostat Energy Balances: nrg_bal_c and nrg_cb_oil |
| Energy sectors | Energy in Mtoe combined with IEA energy prices and Eurostat exchange rates: ert_bil_eur_a |
| Transport sectors | Eurostat Structural Business Statistics: sbs_na_dt_r2 |
| Emissions in Mton of CO2-eq | Eurostat GHG emissions: env_ac_ainah_r2 and env_air_gge |
| Investments matrix and depreciation rates | Eurostat gross fixed capital formation: nama_10_nfa_fl and nama_10_naf_st |
| Employment | Eurostat sectoral employment: nama_10_a64_e |
| Demography | Eurostat population, labour force and unemployment rate: demo_pjanbroad, lfsi_emp_a and une_rt_a |
| Government debt | Eurostat general government gross debt: sdg_17_40 |
Supply and use tables
For France and the Netherlands the SUT data is complete. However, for the EU27 (that is EU27_2020 including Croatia and excluding the United Kingdom) data is only available in basic prices whereas ThreeME assumes purchaser prices. Thus, the product taxes and margins layers have to be estimated.
The only information available on product taxes is one row in the use table containing the total product taxes per sector. The first thing that needs to be estimated is the column in the supply table with total product taxes per product. The 2011 tax rates per product from the EU28 are assumed in order to re-create this missing column. From this estimated column plus the available row in the use table, a RAS-procedure is applied. The initial matrix in the RAS is represented by the intermediate and final use matrix in basic prices. The result is an estimated tax layer by product and sector/final user.
The estimation of the margins layer follows a different approach as no historical data is available at EU level. Therefore, data from other countries is used instead. The average 2015 margin rates by product and sector from twelve EU Member States (AUT, BEL, CZE, DNK, EST, FIN, FRA, HUN, ITA, NLD, PRT and SVK) are used to gap fill the EU27 margin layer. These twelve EU Member States were selected based on data availability.
Physical energy supply and use tables
The ThreeME physical energy layers are provided in Mton and have 20 energy products and 97 sectors (65 NACE sectors plus 28 energy sectors plus four transport sectors). The physical energy data comes directly from the energy balances. The energy balances provide physical energy data by 63 energy products, following the Standard International Energy Products Classification (SIEC), and 112 energy flows.
The 63 SIEC products are aggregated into 20 energy products. The 112 flows in the energy balances are mapped to the ThreeME variables. Some flows are linked directly to the ThreeME variable, like import in energy balances is simply linked to import in ThreeME, while other flows indicate a sector or technology, for instance the final use of energy in agriculture, and are bridged to one of the 97 (NACE) sectors. The correspondence between ThreeME variables and energy balance flows are as follows:
- Domestic energy supply in ThreeME in Mtoe (Y_toe) is assumed to equal the sum of the ‘Primary Production’ and ‘Transformation Output’ from the energy balance. The ‘Primary Production’ flow is assumed to appear only on the diagonal of the monetary supply table. The ‘Transformation Output’ flow can also appear as co-production. Transformation output data is provided by 27 types of technologies. These technologies are bridged to the NACE sectors.
- Intermediate energy consumption in ThreeME in Mtoe (CI_toe) is assumed to equal the sum of the ‘Transformation Input’, ‘Energy Sector’ (excl. auto-consumption), ‘Final Non-Energy Consumption’ and ‘Final Energy Consumption’ flows. All these flows are either given by technology or by NACE sector, which are all bridged to the NACE sectors in the monetary use table.
- Autoconsumption of physical energy in ThreeME (AC_toe) can be found within the ‘Energy Sector’ block from the energy balance. The ‘Energy Sector’ block captures the energy use needed for the extraction or transformation of energy. This includes both the consumption of purchased energy as well as own-produced energy (Eurostat 2019, Energy Balance Guide). The own-produced energy is called autoconsumption and should be excluded from the data in order to keep the consistency with SUTs.
To seperate the auto-consumption from the consumption of purchased energy, the flows in the Energy Sector block are crossed with the flows in the Transformation Output block. If a specific technology uses an energy product and at the same time produces that same energy product, then we can assume that Energy Sector flow is auto-consumption. For instance in France in 2015, coke ovens produce 592 ktoe as transformation output. Coke ovens also consume 20.8 of coke oven gas in the ‘Energy Sector’ block. In this case, the 20.8 ktoe is considered autoconsumption.
Energy exports in Mtoe in ThreeME (X_toe) is assumed to equal the sum of the ‘Exports’, ‘International Maritime Bunkers’ and ‘International Aviation’ flows from the energy balance.
In the energy balances, distribution losses are given per energy product. The distribution losses are assigned to the NACE sectors by assuming a fixed distribution loss rate per product. In ThreeME, the distribution losses from domestic production is separated from distribution losses from imports (DLY_toe and DLM_toe).
In addition, the electricity sectors is further disaggregated to create twelve renewable and non-renewable electricity sectors. The disaggregation shares come from the output of electricity by type of energy source as provided in the last block of the energy balance. This block is a complimentary indicator to the main energy balance table Eurostat (2019).
Lastly, ThreeME requires the distinction between transport diesel and heating gas oil. However, in the energy balance there is just the product called ‘gas/diesel oil’. This product is therefore split into ‘transport diesel’ and ‘heating and other gas oil’ based on commodity balances data. Commodity balances are similar to energy balances but focus on the supply and use of a product rather than transformation efficiencies. The commodity balances are available for a higher level of detail regarding energy products.
Note that a correction is made for nuclear heat which is set to zero. In the energy balances, nuclear heat is considered a primary energy source while in fact it is produced in the reactors IEA (2020). In practice, the primary energy source is nuclear fuel instead of nuclear heat. Nuclear fuels are often imported but this data is not published.
Disaggregation of energy products and sectors
The physical energy supply and use tables, as described in the previous section, are also used to disaggregate energy products and sectors in the monetary supply and use table. First, the physical energy layers are converted to monetary values by multiplying them with the IEA energy prices. The IEA energy prices are provided for industry and consumers as well as excluding and including tax. The corresponding price is used for each element of the supply (excluding taxes) and use table (including taxes). In this way, the monetary energy supply and use is obtained by 20 energy products and 97 sectors.
The values of the 20 energy products and 97 sectors are then incorporated as they are into the monetary SUT. In order to respect the total values from the supply and use table, the energy values are subtracted from their corresponding NACE sector. In some cases the energy value is higher than the value from the NACE sector, which lead to negative values. Negative values are set to zero and the column balance is kept by adjusting the gross operating surplus while the row balance is kept by rescaling all energy values on the demand side.
Disaggregation of rail and road transport
The sector ‘H49 - Land transport and transport via pipelines’ from the SUT data is disaggregated into the following five sectors:
- H491: Passenger rail transport, interurban
- H492: Freight rail transport
- H493: Other passenger land transport
- H494: Freight transport by road and removal services
- H495: Transport via pipeline
Disaggregation shares are taken from Eurostat Structural Business Statistics (SBS). The SBS contain data on different indicators including output, intermediate consumption, value added, employment, wages and employer’s social contributions. Where possible, the disaggregation shares from different indicators are used to disaggregate the corresponding ThreeME variables. However, not all indicators are available for all years and country combinations. In that case, the fall back option is to use just output disaggregation shares for all ThreeME variables.
Investments matrix
In the SUTs, the investments or Gross Fixed Capital Formation (GFCF) are provided by product in one column from the use table. However, ThreeME requires the investments also by sector. Therefore, an investments matrix by product and sector needs to be created.
Eurostat supplies GFCF flows by NACE sector and by asset. This data is assumed to represent the investment matrix. The assets first have to be bridged to CPA2008 products in order to match the SUT classification. Second, the total sum of this matrix should equal the total sum of investments in the use table. For this a RAS-procedure is applied. The inputs of the RAS are:
- row totals that equal the investments in the use table,
- column totals that follow the sectoral shares from the Eurostat GFCF flows
- and the initial matrix represented by the Eurostat GFCF matrix by product and sector.
This procedure results in an estimated investments matrix by product and sector. Note that this approach is only applied for France and the Netherlands as GFCF data for the EU is missing. For the EU27, the intermediate use matrix is used instead of the GCFC matrix, meaning that the investment matrix per product and sector is calculated by applying proportionate shares of the intermediate consumption on the investments per product.
Emissions
In ThreeME, the emissions extensions are not only provided at sectoral level but also by type of fuel as well as by source sector. Eurostat publishes air emissions accounts and air emissions inventories. The difference is that the air emission accounts are fully consistent with the System of National Accounts (SNA) and thus the SUTs whereas the air emission inventories follow international conventions such as the UNFCCC SEEA (2016). The former database is the leading database for ThreeME as ThreeME uses data according to the SNA. The latter database is required as well because of its detailed information on the sources sectors. These source sectors are aggregated in ThreeME into three source sectors that are combustion, production and material related emissions. Production related emissions refer to emissions from agricultural activities as well as the majority of the HFC, PFC, NF3 and SF6 emissions. Material related emissions refer mainly to emissions coming from non-metallic minerals and chemicals.
The distinction of the emissions by type of fuel is made using the information from the energy use table in Mtoe, as estimated in one of the previous steps. This physical use table, provided by 20 energy products and 97 NACE sectors, is converted from Mtoe to Joules using IEA (2005) heat values and from Joules to emissions in tonnes using IPCC (2006) emission factors. These values are then rescaled to match the Eurostat air emissions accounts data.
Redistribution of incomes
The household and government accounts are added to create the fourth quadrant of the Social Accounting Matrix. These include the property income, social benefits, employer social contributions and transfers. Data is used from the Eurostat non-financial transactions. This data also allows for the inclusion of self-employed income (B2A3G: Operating surplus and mixed income, gross) into wages and salaries.