1. IMPACT OF HIGH ENERGY COSTS: RESULTS FROM A GENERAL AND A PARTIAL EQUILIBRIUM MODEL Francesco Gracceva Umberto Ciorba International Energy Workshop Kyoto, 4-6 July 2005

2. 2 OVERVIEW  Energy price and the economy/energy system  Understanding and simulating the link  A general and a partial equilibrium model  Markal-Macro  Markal-ED  Simulation assumptions  Results  Impact of high energy costs on the economy  Impact of high energy costs on the energy system  Conclusions

3. 3 ENERGY PRICES AND THE ECONOMY

4. 4 UNDERSTANDING/SIMULATION OF THE LINK ENERGY COST  ECONOMY/ENERGY SYSTEM (1) MACROECONOMIC EFFECTS Direct effects (more spending for energy costs), depending on:  % energy costs on GDP  dependence on imported energy  ability of the economy to substitute energy with other inputs  … Adjustment effects, linked to real wage, price and structural rigidities:  inflation from increased input costs and pressure on nominal wages  reduced demand and lower investments  risk of unemployment, budget deficit  … Need for a general equilibrium approach

5. 5 UNDERSTANDING/SIMULATION OF THE LINK ENERGY COST  ECONOMY/ENERGY SYSTEM (2) ENERGY SYSTEM EFFECTS Direct effects (energy costs more), depending on:  sensitivity of energy demands to prices and income  ability of the energy system to reduce energy consumption through: conservation efficiency improvements  ability of the energy system to change the fuel mix  … Indirect effects:  through GDP reduction  … Need for a simulation of the endogenous impact of prices/income on demand (together with a detailed description of energy system)

6. 6 A GENERAL AND A PARTIAL EQUILIBRIUM (MARKAL) MODEL

7. 7 MARKAL-MACRO MARKAL MACRO Labor Consumption Energy Costs Energy Investment Capital Y Hydrogen production Industry Residential/ commercial Electricity production Refineries Transport Heating Cooling Power Moving etc. Gasoline Natural gas Electricity Coke Hydrogen Heat etc. Renewables Fossil fuels Nuclear Useful energy Primary energy Conversion sectors/processes Final energy Demand sectors/processes Coke ovens Heat production  Links the technology model MARKAL with a neoclassical economic growth model, through an aggregate (non linear) production function  Variations of energy demand are due to substitution between energy and other production factors, strongly dependent on ESUB  G.E. model: detailed description of the energy sector plus interactions between energy and the economy  BUT, as a single sector model, it is only able to roughly capture many changes in energy demand resulting from changes in exogenous variables

8. 8 MARKAL-ED  An extension of the technology model MARKAL allowing energy service demands to be endogenously affected by its shadow prices and income  The model maximizes net social surplus (based on Equivalence Theorem), a non-linear function linearised by piecewise linear functions  Possible to handle different elasticities for different demand categories  BUT partial equilibrium: no mechanism to simulate macroeconomic feedback effects (even if loss of Net Social Surplus is a proxy of GDP loss) Hydrogen production Industry Residential/ commercial Electricity production Refineries Transport Heating Cooling Power Moving etc. Gasoline Natural gas Electricity Coke Hydrogen Heat etc. Renewables Fossil fuels Nuclear Useful energy Primary energy Conversion sectors/processes Final energy Demand sectors/processes Coke ovens Heat production

9. 9 MARKAL, MARKAL-MACRO, MARKAL-ED Markal Markal-Macro  As most effects are related to measures taken inside the energy sector, “ignoring the impact of GDP on the energy system is not a very significant error” (Loulou, Lavigne, 1996)  Preferable to use: MM to estimate economic effects; models with better simulation of demand sensitivity to prices/income to analyse the impact on the energy sector Modest impact of GDP on energy demand Larger CO2 reduction in MM, mainly due to lower DM/GDP

10. 10 SIMULATING THE LINK BETWEEN ENERGY COST AND THE ECONOMY/ENERGY SYSTEM: A TWO MODEL APPROACH MARKAL-ITALY Evolution of energy system up to 2048 Equilibrium quantities and prices of more than 300 flows of energy goods and materials, from more than one thousand technologies 67 energy service demands:  Industry: production of energy-intensive materials  Residential: heating, cooling, lighting, warm water, main electric devices  Services: heating, cooling, lighting, other electric use and motive power  Transportation: pass-km and ton-km, three areas (urban, medium and long distance)  Agriculture Energy system effects Markal-Macro Italy Markal-ED Italy (price+income elastic) Assessment of economic effects

11. 11 SIMULATION ASSUMPTIONS

12. 12 THE REFERENCE SCENARIO  Socio-economic assumptions GDP (average growth) Population (avg. gr.) 1.8% -0.1% 1.8% -0.1% 2004-20122012–20202020-2028 1.8% -0.2%  Price of imported fuels

13. 13 OIL PRICE IN REFERENCE AND HIGH PRICE SCENARIOS Two scenarios: difference a bout 10$/bbl

14. 14 ASSUMPTIONS ABOUT SOME KEY MODEL PARAMETERS  Price elasticities Industry-0.5 Services-0.4 Residential-0.3 Transp. - passengers-0.4 Transp. - freights-0.2  Income elasticities Industry0.8 Services0.8 Residential0.6-0.8 Transp. - passengers1.0 Transp. - freights0.8-1.2 MARKAL-MACRO:  ESUB (elasticity of substitution between capital/labour and energy) = 0.35  GROWV (potential annual GDP growth rates) = 1.8%/yr MARKAL-ED:  Maximum demand reduction/increase = from 20% in 2004 to 50% in 2028  Alternative GDP growth = same GDP growth projected by in MM-High Price

15. 15 RESULTS

16. 16 HIGH ENERGY PRICES AND THE ECONOMY  The economic impact of high energy price is significant (-0.2-0.3%). Estimates are similar to recent literature (IEA, 2004)  The economic effect of high prices is quite similar according to the two versions of the model: loss of Net Social Surplus in MED is similar to GDP loss in MM

17. 17 HIGH PRICES AND THE ENERGY SYSTEM (1/5): MARKAL-ED vs. MARKAL-MACRO  The impact of high prices on the energy system is quite different: reduction of energy service demands is significantly larger in MM  Introduction of income elasticity in partial equilibrium model increases the reduction of demands only slightly (in medium-term)

18. 18 HIGH PRICES AND THE ENERGY SYSTEM (2/5): MARKAL-ED vs. MARKAL-MACRO  The two versions of the model produce quite different reductions  In MM, strongly dependent by a key parameter (esub), reduction in almost all sectors  A result produced by different methodology or by different parameters ?

19. 19 HIGH PRICES AND THE ENERGY SYSTEM (3/5) Markal-Macro:  Reduction of final consumption increases with time  In medium/long-term reduction reaches about -2 Mtoe (mainly in industry and residential) Markal-ED:  Reduction of final consumption in industry follows reduction in service demand  Reduction of final consumption in other sectors follows long- term efficiency improvement

20. 20 HIGH PRICES AND THE ENERGY SYSTEM (4/5)  Change in fuel mix: accelerated substitution between oil and gas (with respect to Reference)

21. 21 HIGH PRICES AND THE ENERGY SYSTEM (5/5) Markal-Macro: CO2 reduction is strong in  industry (demand reduction)  electricity sector (change in fuel mix) Markal-ED: CO2 reduction is large mainly in industry, the only sector with significant demand reduction

22. 22 CONCLUSIONS/OPEN ISSUES GE approach is useful to estimate the impact of energy prices/measures on the economy, but GE models usually simulate only roughly the link between energy cost and energy demand  Effects on the energy system overestimated by GE models ? On the contrary, impact on the energy system is better simulated with more detailed simulation of the sensitivity of energy demands to price/income  (Provided that good elasticity estimates are available), partial equilibrium models are preferable to analyse impact on energy system ? Some results from MM-Italy+MED-Italy:  economic impact of high energy price is significant, and estimates are similar to recent literature  the impact on the energy sector is quite larger in MM than in MED  A possible relevant policy implication: even if economic effects are significant, expected reduction in energy consumption (and emissions) could be overestimated