The best of two traditions: Integrating bottom-up information in CGE models, including TIMES input

Size: px
Start display at page:

Download "The best of two traditions: Integrating bottom-up information in CGE models, including TIMES input"

Transcription

1 Taran Fæhn, Research Dep., Statistics Norway The best of two traditions: Integrating bottom-up information in CGE models, including TIMES input Linking CGE and TIMES Models - Workshop Technology and Innovation Centre, Strathclyde Univ., 09 Nov

2 Background Traditional approaches overestimate the costs of climate policies: potential abatement options are omitted TOP-DOWN approach (e.g. CGE-models) Disregard most opportunities for future technologies Assume technologies of today (when calibrating) and of yesterday (when estimating) BOTTOM-UP approach (e.g. energy system models): Disregard most reallocations to cleaner activities taking place when emitting is costly Exogenous consumption and production patterns The approaches complement each other and should be combined 2

3 Analyses of climate and energy policies with linking-approaches Soft-linking CGE and MARKAL (Bjertnæs, Martinsen&Tsygankova,2013; Martinsen, 2011) Integrating bottom-up info in CGE 1: Inserting marginal abatement cost functions (Fæhn&Isaksen, 2016) 2: Abatement as a CES composite (Bye, Fæhn&Rosnes, 2015) Further plans 3

4 Soft-linking CGE and MARKAL a) Bjertnæs, Martinsen&Tsygankova (2013, Energy Economics 39) b) Martinsen (2011, Energy Policy 39) Research questions: Effects of unilateral vs. global carbon pricing on a) public budgets, welfare and emissions b) learning and costs of technologies Approach: Three models in cooperation 1) Global MARKAL IN: carbon pricing OUT: Global energy prices, learning effects no feedback 2) National MARKAL IN: energy prices + technology costs from 1); demand from 2) OUT: system costs and emissions 3) National CGE IN: sector-dispersed system costs (acc. to sectorial pattern, as productivity loss) and total emission cuts (added to those from CGE), OUT: public budgets, welfare 4

5 Soft-linking CGE and MARKAL Some lessons learned: We gained in terms of: more response possibilities and related costs interaction between national and global responses The costs were large: the models have different aggregation matching sectors and instruments much overlapping and inconsistent endogeneity iteration was costly ex ante (communicating across disciplines and trial and error with technical solutions) iterating model solutions was also costly (many simulations) The result: not fully iterated solutions two different publications with focus on each model/discipline 5

6 Costs Integrating bottom-up info in CGE Three different approaches: 1) Inserting estimated sector-wise abatement equations in CGE (1) (2) (3) (4) (5) (6) D f ( c) U / U.0 C cd ~ U U D U / X ~ V V C X / V f ( c) dc U / U.0 (Fæhn and Isaksen, Energy Journal, 2015) 2) Introducing abatement as a CES composite of emissions and capital (Kiuila and Rutherford, Energy efficiency: Ecological Economics, 2013; Abatement Bye et al., SSB-DP 817/2016) Emissions Capital 3) Introduce each abatement option as Leontief functions - activate the profitable step by step (Böhringer, Energy Economics,1998) Abatement by technology 6

7 1) Inserting estimated sector-wise abatement equations in CGE Fæhn and Isaksen (2016, Energy Journal 37/2) Research question: What if commitment problems hamper investments in clean technologies? Carbon pricing and subsidies to reach national targets Approach: 1) Bottom-up data on sector-wise technology options included as MACs 2) Technological abatement added to the sector s abatement through CGE reallocations and factor substitution Results: Without confidence in persistent policy, no investments and a trippling of abatement costs. Up-front subsidies is a second-best policy option. 7

8 /t CO2e 1) Inserting estimated sector-wise abatement equations in CGE Bottom-up input information (Process industry): Abatement measure Annuity (EUR/tCO 2 e) Abatement (Mt CO 2 e) Accumulated abatement (Mt CO 2 e) a Process optimisation (metals) b Energy efficiency and substitution (metals) C Energy efficiency and substitution (pulp and paper) d Substitution of bio (cement and other minerals) e Energy efficiency and substitution (chemicals) f <40% charcoal for coke (ferrosilicon) g <20% charcoal for coke (ferromanganese) h <80% charcoal for coke (ferrosilicon) i Substitution of bio (cement) j Process optimisation (petrochemicals) k Charcoal substitute for coke (silicon carbide) l Substitution of bio (anodes) m CCS (fertilisers) n CCS (cement) o 500 Substitution of bio (pulp and paper) Mt CO2e 8

9 /t CO2e 1) Inserting estimated sector-wise abatement equations in CGE Bottom-up input information (+ Petroleum extraction): Abatement measure Annuity Abatement (million Accumulated abatement (EUR/tonne CO 2 e) tonnes CO 2 e) (million tonnes CO 2 e) a Energy efficiency offshore b Electrification Melkøya c Electrification Melkøya d Electrification Melkøya e Mongstad processing CCS f Electrification North Sea south G Electrification new site h Electrification North Sea north i Kårstø processing CCS petroleum process Mt CO2e 9

10 /t CO2e 1) Inserting estimated sector-wise abatement equations in CGE Bottom-up input information (+ Road transport): Abatement measure Annuity (EUR/ tonne CO 2 e) Abatement (million tonnes CO 2 e) Accumulated abatement (million tonnes CO 2 e) a Efficiency improvements private cars level b Efficiency improvements private cars level c Zero emissions vehicles private and public d Intermixture of ethanol E e Intermixture of 1.generation biodiesel f Intermixture of ethanol E5, E10, E g Intermixture of 2. generation biodiesel Fig process 2.1 petroleum Fig 2.2 transport Fig Mt CO2e 10

11 /t CO2e 1) Inserting estimated sector-wise abatement equations in CGE Estimated marginal abatement cost (MAC) curves: 500 y = 13,31x 3-35,547x ,021x R² = 0,9829 y = 15,076x 3-99,501x ,12x R² = 0, y = 7,843x ,851x R² = 0,

12 1) Inserting estimated sector-wise abatement equations in CGE The CGE adjustments: 1) Technological abatement curve= Relationship between accumulated abatement and marginal costs (x scale factor) 2) Total abatement = abatement in the traditional model+resulting from climate technology investm. 3) Endogenous em.coefficients 4) Total abatement costs= integral above curve = added input costs in the industry (less efficient inputs) 5) Total production costs include the abatement costs 6) Endogenous productivity The equations: (1) (2) ~ ~ D f ( c) U / U.0 ~ U U D (3) U / X (4) (5) (6) C V cd ~ V C X / V f ( c) dc U / U.0 12

13 1) Inserting estimated sector-wise abatement equations in CGE Some lessons learned We gained in terms of: Sector-wise technology information could be exploited While soft-linking procedures need to be repeated for every project and simulation, integration is made once and for all Still potential for improvements: Abatement technologies assumed to have same factor intensities as the production technology -> Wrong factor-market responses Potential double counting, as estimated substitution elasticities may embody abatement potentials (e.g. between energy goods or between energy and capital ) 13

14 2)Introducing abatement as a CES composite of emissions and capital Bye, Fæhn&Rosnes (2015, SSB-DP 817) Research question: Introducing energy efficiency targets in households effects on household behaviour, rebound, emissions and welfare. Approach: 1) Limited to energy efficiency measures in households. Bottom-up data from the TIMES model. 2) Used the data to estimate substitutiton elasticity between energy and capital (3) Analogue procedure can be used between emissions and capital) Results: Energy efficiency gains come at a substantial cost and also with 40% energy rebound to the economy at large. 14

15 2)Introducing energy efficiency improvements as a CES composite of energy and capital The CES structure of household consumption: Consumption Housing Transport Other goods and services Dwelling Energy Transport Fuel 1 n Electricity Fossil Gas Paraffin and heating oil Fuel wood, coal, etc. District

16 2)Introducing energy efficiency improvements as a CES composite of energy and capital Available energy efficiency measure information from TIMES-Norway Household Measure Investment cost Lifetime Saving potential (Gwh/yr) Annuities Excess cost Type code type /kwh yrs 2010 /kwh % Existing single unit RSIOH031 Post-insulation roof 2, , % New blocks RMUNH045 Solar collector 2, , % New single unit RSINH045 Solar collectors, new 2, , % Existing blocks RMUOH033 Post-insulation wall 3, , % Existing blocks RMUOH032 Post-insulation floor 3, , % Existing single unit RSIOH032 Post-insulation floor 3, , % Existing blocks RMUOH045 Solar collectors, rehab 3, , % Existing single unit RSIOH045 Solar collectors, rehab 3, , % Existing single unit RSIOH033 Post-insulation wall 4, , % Existing blocks RMUOH034 Replacement entrance door 4, , % Existing single unit RSIOH034 Replacement entrance door 4, , % Existing blocks RMUOH035 Replacement windows 10, , % Existing single unit RSIOH035 Replacement windows 10, , % 16

17 Relative investment costs 6 2)Introducing energy efficiency improvements as a CES composite of energy and capital Investment costs of energy savings data and calibrated substitution DATA ESTIMATED CES (0.3) Energy savings base year, TWh 17

18 2)Introducing energy efficiency improvements as a CES composite of energy and capital Some lessons learned We gained in terms of: The new elasticity is based on future expectations, not old, outdated? information. Much technology information abstracted to one parameter Still potential for improvements: Not all technology information is relevantly captured by the substitution elasticity more parameters could be calibrated anaogously to capture more Technology data typically include measures with negative costs need to identify the reason for such data to treat them correctly The measures cannot be precisely identified in the CGE results extrapolations Smoothed curves costs typically have discrete jumps, not incremental steps New information requires new estimations/calibrations 18

19 Costs Further plans Combine the two approaches: Energy efficiency and energy mix changes better represented by substitution elasticities, Process emissions and end-of-the-pipe-like abatement better represented by MACs We avoid double-counting Use stepwise MAC functions Measures can be precisely identified Don t need any estimations easier to update Input-output information of the measures are represented and included in the input-output system of the CGE right factor-market responses Abatement by technology 19

20 Summing up the experiences in SSB Soft-linking top-down and bottom-up models (Bjertnæs,Martinsen&Tsygankova 2013, Martinsen 2011) Doable, but inconsistencies remain Time-consuming and must be repeated for every new policy case Integrating technological information into CGE (Fæhn&Isaksen, 2016; Bye, Fæhn&Rosnes, 2015) By far the most promising - use bottom-up model inputs instead of outputs - integrate in CGE model once and for all 20

21 Thank you for the attention REFERENCES: Bjertnæs, G, T. Martinsen, M. Rybalka (2013): Norwegian climate policy reforms in the presence of an international quota market, Energy Economics 39, Martinsen, T (2011): Introducing technology learning for energy technologies in a national CGE model through softlinks to global and national energy models, Energy Policy 39, Fæhn, T. and E.T. Isaksen (2016): Diffusion of climate technologies in the presence of commitment problems, Energy Journal 37 (2), Bye, B., T. Fæhn, O. Rosnes (2015): Residential energy efficiency and European carbon policies: A CGE-analysis with bottom-up information on energy efficiency technologies, Discussion Papers No. 817, Statistics Norway. 21