Estimating Potential Demand for Electric Vehicles EVs Michael

Estimating Potential Demand for Electric Vehicles (EVs) Michael K. Hidrue and George R. Parsons Camp Resources XVII Wrightsville Beach, NC June 24 -25, 2010 Sponsored by: US Department of Energy, Office of Electricity Delivery and Reliability

Outline • • • Objective Study design Estimation results WTP estimates Conclusion 1

Objectives • Estimate potential market demand for EVs • Assess the value of adding V 2 G on demand for EVs • V 2 G vehicles are special type of EVs that allow people to sell power from their batteries back to electric companies. 2

Study Design • Web based choice experiment • National Survey, N=3029 • Sample resembles national census data • Latent class random utility model 3

Sample EV Choice Set 4

Results: Number of Latent Classes • BIC identified two latent classes EV class GV class • The EV class has positive EV constant, high value for fuel saving and tend to be green • The GV class has negative EV constant, low value for fuel cost saving and tend not to be green 5

Results: Class Membership Model Variable Coefficient T-stat Odds ratio Constant Young (25 -35 yrs old) -2. 9 -12 6. 2 0. 06 Middle age (36 -55 yrs old) 0. 26 2. 4 1. 3 Expected gas price in 5 years 2. 6 1. 07 Change in life style & shopping habit : Major 0. 07 1. 15 7. 8 3. 2 Change in life style & shopping habit: Minor 0. 71 5. 8 2. 0 Having access for installing charger 1. 14 10. 1 3. 1 Expected next vehicle: Hybrid 1. 06 10. 4 2. 9 Multicar household -0. 12 -1. 2 0. 9 College (>=B. A) 0. 06 0. 7 1. 1 Male 0. 34 3. 7 1. 4 0. 78 2. 2 Four more variables… 6

Results: Vehicle Choice Model Parameters Attributes GV class EV class Coef. T-stat. EV constant -4. 02 -10. 0 0. 47 4. 0 Yea saying -0. 08 -0. 27 -5. 0 Price -1. 8 E-04 -6. 0 -1. 04 E-04 -25. 1 Pr *pr on GV 4. 1 E-07 0. 04 1. 1 E-06 6. 9 Fuel cost -0. 16 -1. 4 -0. 34 -9. 3 Prob. Weighted Implicit Prices -$9, 337* -$2, 764 *=value of yea saying is subtracted from the constants 7

Results: Vehicle Choice Model Continued Parameters Attributes GV class Coef. EV class T-stat. Coef. T-stat. Probability weighted Implicit Prices Driving range (Ref=75 mi) 150 mi 0. 81 3. 8 0. 44 7. 3 $5, 177 200 mi 0. 89 4. 3 0. 84 14. 6 $8, 134 300 mi 1. 34 6. 4 1. 14 17. 4 $11, 391 Charging time for 50 mi (Ref=10 hrs) 5 hours 0. 57 2. 7 0. 09 1. 6 $2, 136 1 hour 1. 04 5. 3 0. 45 8. 3 $5, 858 10 minutes 1. 31 6. 8 0. 74 13. 6 $8, 567 8

Results: Vehicle Choice Model Continued Attributes Parameters GV class Coef. EV class T-stat. Coef. T-stat. Probability weighted Implicit Prices Acceleration relative to respondent’s next car (Ref=20% slower) 5% slower 0. 58 2. 8 0. 06 1. 04 5% faster 0. 92 4. 2 0. 28 4. 7 $4, 372 20% faster 1. 16 5. 3 0. 50 8. 2 $6, 521 $1, 957 Pollution relative to respondent’s next car (Ref=25% lower) 50% lower 0. 11 0. 52 0. 05 0. 84 $636 75% lower 0. 33 1. 9 0. 08 1. 3 $1, 428 95% lower 0. 53 2. 8 0. 28 4. 8 $3, 333 9

Top 10% WTP Estimates Assumptions: Fuel cost=$1. 00/gal equivalent Acceleration=5% slower Pollution=75% lower 10

Comparing WTP Estimates with Battery Cost Estimates 11

Comparing WTP and Battery Cost Estimates 12

Comparing WTP and Battery Cost Estimates in the Presence of Subsidy 13

Conclusion • Driving range, charging time and performance are significant drivers of EV choice • Green life style, hybrid buyer, outlet access, expected gas price and age are significant predictors of EV choice • Multicar household, college education and regions are not significant predictors of EV choice • People will pay premium for some EV designs • For EVs to compete on the market, battery cost has to decline substantially. 14