1. Modelling the energy demand of households Kurt Kratena, Ina Meyer, Michael Wueger WIFO (Austrian Institute of Economic Research)

2. Demand system with stocks and ‚service prices‘ Expenditure function for non-durables with utility (u) and prices (p i ) + Expenditure for durables I with price (p I ) Main features: -Appliance stocks  energy efficiency -Dealing with “service prices” and demand (‘rebound’ effect !) -Combining time series & cross section estimation

3. Demand system with stocks and ‚service prices‘ Converting energy flow (E) into service (S): Impact of the efficiency parameter (  ES ) on the ‘real price of service’ Budget shares = service shares

4. Demand system with stocks and ‚service prices‘ Optimality conditions for cost minimizing: Shephard’s Lemmaenvelope condition Impact of capital stock on expenditure given by efficiency improvement-effect: : technical progress & consumers’ choice

5. Impact of stock (changes) on efficiency : technical progress & consumers’ choice ADL-model with long run elasticity of efficiency wrt. to energy prices (consumers choice) and capital stock (autonomous/embodied technical progress) Elasticities:

6. Almost Ideal Demand System (AIDS) or Quadratic AIDS for C(u, p i ) Budget share of AIDS Budget share of QUAIDS Restrictions

7. AIDS model for C(u, p i ) Elasticities 1. Income elasticities (direct derivation) 2. Price elasticities General:  ij,COMP =  ij,UNCOMP +  i w j (Slutsky equation)  ij,COMP …compensated price elasticity  ij,UNCOMP …uncompensated price elasticity.

8. AIDS model for C(u, p i ) Elasticities 2a. Uncompensated price elasticity where  ij is the Kronecker delta and  ij = 1 for i = j and  ij = 0 for i ≠ j. 2b. Compensated price elasticity

9. Quadratic AIDS model for C(u, p i ) Elasticities 1. Differentiate w i wrt. to C and p j :

10. The Quadratic AIDS model Elasticities 2. Derive elasticities from  ’s : Income elasticity Uncompensated price elasticity where  ij is the Kronecker delta and  ij = 1 for i = j and  ij = 0 for i ≠ j. General:  ij,COMP =  ij,UNCOMP +  i w j (Slutsky equation)

11. Dynamics of energy demand (E i ) Totally differentiating E wrt. time (t) Direct effect & indirect effect via service demand. Total impact: Direct (price induced-price) rebound effect Indirect (price induced-income) rebound effect

12. Empirical application Austria (1990 – 2006) USA (1972 – 2005) Austria (1990 – 2006) & Household Budget Survey 2004/05 - National Accounts (private consumption, COICOP) -Efficiency of household appliances: Refrigerators, freezers, washing machines, dish washers, TVs, dryers, heating, water heating and cooking (National Lawrence Berkeley Laboratory, ODYSSEE database) -Efficiency of private car fleet

13. Efficiency, Austria

16. Efficiency, U.S.

17. Efficiency, U.S.: energy & service price

20. Estimation results: price elasticities, Austria

21. Empirical results: Austria Rebound effects: Gasoline: 50%, heating fuels: 20%, electricity: 10%, range in the literature: 10% - 30%. Pure income rebound effects: Gasoline: -1.9%, heating fuels: 6.6%, electricity: 2%. Decomposing energy demand: Further decomposition of dS/S into: price rebound, income rebound, other factors

22. Decomposition: Austria

23. Estimation results: price elasticities, U.S.

24. Empirical results: U.S. Rebound effects: Gasoline: 16%, heating fuels: 21%, electricity: 8%, range in the literature: 10% - 30%. Pure income rebound effects: electricity 9.4 %. Long-run elasticity of efficiency

25. Simulation results: U.S. (long run change in energy prices and capital stocks)

26. Conclusions Rebound effects: main link between top down and bottom up modelling Efficiency has only a limited impact on energy demand during low energy price-periods  service demand is the driver of energy demand Long run impact of prices exceeds short run impact by far