Deriving the carbon footprint

We combine energy IO analysis with emission factors and household expenditures for goods and services to estimate the carbon footprint for Indian households. Therewith, we can trace the carbon content of each final consumption item back to its intermediates and account for the direct as well as indirect emissions from consumption. We focus on carbon emissions from fossil fuels since CO₂emissions represent the largest share of GHG emissions and other GHG emissions are not available on the used detailed sectoral level. The method which has been applied is based on Leontief (1970) and we follow the approach of Proops et al. (1993) and Lenzen (1998b) summarized in Munksgaard et al. (2009). In a first step we estimate the CO₂intensities (in local currency unit) of each sector of the Indian economy. We apply a single region IO model based on data from the Global Trade Analysis Project (GTAP).

By using a single region IO model we account for direct and indirect emissions from goods produced and consumed in India as well as for emissions from imported goods.

IO tables for the year 2004 are from the Indian Central Statistical Organisation (CSO) which provide us with an [j×1] vector of domestic output x by 130 sectors j, a [j×1]

2.3 Methodology 21 vector of final demandyby 130 sectors j(which includes imports).² TheA[j×j]matrix of the technical coefficientsAreflect the input requirements of the jth sector of intermediates from other sectors measured in monetary units.³ The domestic technology assumption is applied with the assumption that imported goods are produced with the same technology as local goods. We also assume that technology has not changed drastically between 2004 and 2011 since we use the same IO table to estimate the emission intensities of sectors for 2011.⁴ Depending on the fuel type the CO₂emissions per unit of fuel use are represented in the emission coefficient vectorc[m×1]. The[m×j]energy use matrixE_ind represents the quantitative energy demand of the 58 sectors per monetary unit of intermediate output from other sectors. The energy use matrixE_{f d} represents the household’s fuel quantitative use per monetary unit of final demand from 58 sectors.⁵ Total emissions from consumption CO₂would consist of directCO2_{f d} from final demand and indirectCO2_ind emissions from energy use by each sector. In the first step we match the 130 sectors of our IO tables with the energy use data, which is aggregated to 58 sectors in order to get the energy intensity matrixE. Secondly, we match the 58 sector emission intensities with the corresponding expenditure categories from the household survey data. The data on household expenditure is rather disaggregated and we match all the approximately 340 expenditure categories with the corresponding emission intensities. Even though the IO tables contain information on monetary fossil fuel and electricity demand we still need to refer to the quantitative energy intensity data from GTAP to gain a more precise estimate on emissions per sector.

We analyze the sum of direct and indirect emissions from industrial sectors. Direct emissions from final demand can be characterized as follows:

CO2_{f d} =c^′E_{f d}y (2.1)

wherec^′represents the inverse emissions coefficient vector,E_{f d} is the energy use matrix and yis the final demand vector. Indirect emissionsCO2_ind, which are divided into emissions from domestic production for domestic final demand, emissions from imported intermediates and emissions from imported final demand.⁶ The emissions by sector can be estimated by multiplying the demand of each sector represented as vectorywith the transposed emissions coefficients vectorcand the industrial energy use matrixE_indas well as the with the domestic

2The 130 sectors include administration and defence.

3All values are in local currency units at 2004 producer prices.

4This assumption is confirmed by the emission intensities per sector from the World Input Output Database (WIOD), which did not change drastically in India for available years between 2004 and 2009

5The data by the GTAP energy volume data is disaggregated into 58 sectors, which were matched with the 130 sectors from the Indian IO tables.

6Exports are excluded.

22 The carbon footprint of Indian households identity matrix andAis the technical coefficients matrix, which mirrors the contribution of the intermediates to one final output unit. Additional to these direct and indirect production emissions, households directly use fuels which are unaccounted for in the IO analysis. With observed quantities in the survey data, we calculate direct carbon intensities for kerosene, liquefied petroleum gas (LPG), petrol and diesel.⁷ Another correction is necessary for electricity emission intensities, due to the block-tariff nature of electricity expenditures.

Since prices per kilowatt hour rise with the usage of electricity, we find no constant but household specific carbon intensities. Lower income households with small electricity demand have higher carbon intensities per currency unit than higher income households with large demand. The calculation is based on observed quantities of electricity demand.⁸

Direct emissions from fuel use CO2_{f d} and direct and indirect production emissions CO2_ind embedded in consumption (with the exception of electricity)can be estimated by:

CO2=c^′h

E_{f d}y_hh+E_ind

(I−A_tot)⁻¹y_̸=exp+ ((I−A_tot)⁻¹−(I−A)⁻¹)y_̸=exp

+ (I−A_tot)⁻¹y_imp̸=expi (2.3)

In order to estimate the household carbon footprint we multiply the carbon intensity σ =c^′[E_{f d}+E_ind(I−A)⁻¹]per local currency unit of each industrial sector with the house-hold expenditure for the respective category and sum up over all consumption categories for each household. Therewith we gain the household carbon footprintCO2_hh for each household in tonnes (t) ofCO2.

CO2_hh=

n

∑

j=1

(σ_j∗exp_j) (2.4)

whereirepresents the household and jthe different expenditure category. The household expenditure data for the carbon footprint calculation is from the National Sample Survey Organisation (NSSO) India, with a sample size of approximately 125000 and 100000

house-7Direct carbon contents are 0.00255 t/l (kerosene), 0.00159 t/kg(LPG), 0.00231 t/l (petrol) and 0.00273 t/l (diesel).

8Electricity production carbon intensity is taken as 0.001003 t/kwh (IPCC, 2005)

2.3 Methodology 23 holds in 2004/05 (61st round) and 2011/12 (68th round) respectively. Household expenditures are disaggregated into around 346 consumption items which we aggregate to 19 expenditure groups (for a description of expenditure groups see table 2.1 and the matching scheme with CSO IO tables for 2011/12, see table A.1).⁹