1. Household Energy Use and Travel: Opportunities for Behavioral Change Sashank Musti, Katherine Kortum and Dr. Kara Kockelman The University of Texas at Austin

2. Questionnaire Design Cover letter (English and Spanish) Five sections –Travel Choices –Vehicle Ownership –Home Design and Energy Use –Energy Policy Opinions –Demographics

3. Survey Distribution Westlake Sunset Valley East Austin Manor Hyde Park Far West

4. Survey Distribution (2) Central Market Grocery Flyers and URL cards Community organizations Web links via CapMetro and City sites Internet version of the survey: www.energysurvey.co.nr www.energysurvey.co.nr

5. Data Weighting Sample was compared to PUMS Six control attributes: 720 categories –Gender (male, female) –Student status (student, non-student) –Worker status (worker, non-worker) –Age (18-24, 25-34, 35-44, 45-54, 55-64, 65+) –Household Size (1, 2, 3, 4, 5+) –Income ( $75k) Categories with few observations combined

6. Sample vs. Austin Workers are under-represented (nearly 2 to 1). Students are very over-represented. VariableSampleAustin Female49.6%50.4% Age 45+37.5%38.7% High-income46.6%39.5% Employed37%70.3% Students82%13%

7. What Should We Do?

8. Where Do We Stand?

9. Yearly VMT per Person (WLS) VariableCoefficientT-statistic Mean Elasticity Constant59035.30- College-educated-1437-2.67-0.135 Income per person0.02031.770.099 Number of children-1385-5.48-0.071 Distance to CBD307.46.060.2254 Population density124.61.710.069 Population density & distance to CBD >6 -209.49-1.68-0.023 Transit stops-16.34-2.18-0.054 Age of respondent25.461.460.127 R2R2 0.1289 Adjusted R 2 0.1191

10. Yearly Fuel Use per Person (WLS) VariableCoefficientT-statistic Mean Elasticity Constant388.186.42- College-educated-101.18-3.00-0.170 Income per person9.25 E-041.300.080 Number of children-71.17-4.60-0.065 Distance to CBD16.65.080.218 Population density & distance to CBD >6 -11.04-1.51-0.022 Transit stops-1.17-2.43-0.069 Age of respondent1.581.500.140 R2R2 0.1263 Adjusted R 2 0.1169

11. Yearly VMT and Fuel Use Both increase as… –Distance to CBD increases –Age increases Both decrease as… –Education level rises –Number of children increases –Number of transit stops increases

12. Home Size and Monthly kWh (WLS) SQUARE FOOTAGEMONTHLY ELECTRICITY Independent VariablesCoefficientT-statistic Mean Elasticity CoefficientT-statistic Mean Elasticity Constant10068.28-701.94.15- Household size44.161.550.066977.302.950.1109 Worker status143.11.800.0316--- Income ($1,000)3.9 E-036.390.18971.16 E-031.130.0540 College-educated-221.8-3.72-0.097--- Age of home-9.09-5.77-0.17542.701.050.0491 Own home424.55.410.1654-186.7-1.42-0.0688 Number of vehicles222.75.280.1946--- Number of adults65.151.980.0849--- Job density-1.718-1.83-0.0077--- Population density-21.19-2.94-0.0544-25.05-0.98-0.0608 Two- & three-story detached home ---355.23.260.05 Home size---0.49186.460.4687 R2R2 0.36460.1958 Adjusted R 2 0.35420.1844

13. Home Size and Monthly kWh (WLS) Both increase as… –Income increases –Household size increases Both decrease as… –The area grows denser Older homes tend to be smaller but use more electricity. College graduates tend to have smaller home sizes.

14. Comparison to EIA’s RECS Data ComparisonAustin Energy Survey RECS, 2001 Average home size1,645 sq.ft.2,100 sq.ft. Average monthly kWh1,200 kWh900 kWh +1 household member+ 77 kWh+ 104 kWh +100 square feet+ 49 kWh+ 22 kWh

15. Opinions on Climate Change (Binary Probit) REGULATIONS SHOULD BE IMPOSED ADAPT TO A WARMER CLIMATE Explanatory VariablesCoefficientT-statisticCoefficientT-statistic Number of household vehicles-0.1136-1.730.25714.04 Age of respondent-0.0189-5.20-- Female0.31923.54-0.4526-4.76 Worker status-0.3136-2.360.34053.03 Middle income indicator (between $80,000 & $112,500) 0.32672.83-- College-educated--0.22172.13 Own home0.15421.26-- Rooms in home-0.0729-3.44-- Age of home0.0763.200.0994.03 Average annual VMT---1.80 E-05-3.18 Constant1.03214.65-1.1067-6.03 Log Likelihood at Convergence -540.37-473.94 Pseudo R 2 0.072510.07410

16. Opinions on Climate Change (Binary Probit) Regulations preferred by… –Women –Homeowners Adaptation preferred by… –Workers –Households with many vehicles Those with older homes acknowledge the importance of both regulations and adaptation.

17. Energy Reduction Strategies (Bivariate Ordered Probit) CAP ON ENERGY USETAX ALL ENERGY USE Explanatory VariablesCoefficientT-statisticCoefficientT-statistic Number of household vehicles0.12731.45-0.2350-2.74 Age of respondent0.02113.85-- Female-0.1196-1.030.22301.86 Number of workers-0.1495-2.27-- College-educated--0.24961.92 Income per household member---4.02 E-06-1.51 Household size--0.058291.89 Worker status0.40822.69-- VMT per household member-1.19 E-05-1.402.9 E-052.96 Age of home-0.076-2.43-0.048-1.76 Rooms in home0.08552.56-- Own home-0.1745-1.210.16421.10 Threshold 1-0.2276-0.841-1.4378-6.32 Threshold 20.93323.55-0.1253-0.552 Threshold 31.3855.140.41491.89 Threshold 42.1047.841.2095.42 Log Likelihood at Convergence-2079.4015 Covariance across equations’ residuals0.21243.37

18. Energy Reduction Strategies CAPPING is preferred by… –Households with many vehicles –Older respondents –Workers TAXATION is preferred by… –College graduates –Large households –Homeowners

19. Conclusions Long-term behavioral changes are difficult to implement. Most agree climate change is a concern, but are unwilling to change their own behavior. Increasing income and education lead to greater (stated) concern about one’s impact on the environment.

20. Conclusions (2) Electricity usage increases by 77 kWh/month for an additional person in a household & by 49 kWh/month for an additional 100 square feet of living space. Average electricity consumption can be reduced by moving into newer, smaller homes. Fuel consumption increases by 16.6 gallon/person with a one mile increase in driving distance to the CBD. VMT per person per year increases by 307 miles with every additional mile a household lives from the CBD.

21. Thank You for your attention. Questions and Suggestions?

22. Sashank needs to rename & recluster/list variables, & get elasticities alongside, but there is enough info in these results for Katherine to start inferring meaningful results & she has done a nice job of that in the ppt she sent. (E.g., what's most pract signif/relevant, what is not in there that you thought you'd see, & *how can OTHERS make use of these results* ;-) Katherine: I'm afraid 22 slides is probably too much for the allotted time (I think it would take me close to 20 min., & you should aim for 18. You can tell how long it takes by practicing slowly.) I'm in room 2415 in case you want to try & reach us. I'd highlight 2 to 3 variables you'd like to talk about in a table. (Also, I think you could get away from such a slide altogether, though I do like how you accomplish/review two models with a single slide, as with slide 15. E.g., you could highlight 2 or 3 that signif increase the response with "red" [bad things, increasing c02/energy, for example], and the good practically significant variables [the ones that reduce vmt/fuel] with green.) On slide 19, what I think is interesting is who supports the first & not the 2nd (e.g., females & non-workers, in hh's with fewer vehicles). The college educated may simply realize that there will have to be adaptation (i.e., climate change is here to stay).

23. Vehicles per Household (Poisson) Independent Variables CoefficientT-statistic Household income3.93 E-069.13 Income per person-4.42 E-06-4.49 Number of workers in the household-0.0448-2.64 Own home0.27435.28 Annual VMT1.16 E-057.69 Age of respondent0.02801.71 Region-Specific Variables Distance to UT campus0.02401.49 Household density-0.0564-1.59 Population density0.02591.66 Job density-2.07 E-03-1.61 Constant-0.2690-3.09

24. Vehicles per Household (Poisson) Comments (the model included in this presentation is currently an old model; among other things, it includes two income variables)

25. Energy Reduction Strategies (Ordered Probit) CAP ON ENERGY USETAX ALL ENERGY USE Explanatory Variables CoefficientT-statisticCoefficientT-statistic Number of household vehicles0.12461.42-0.2417-3.94 Age of respondent0.019923.59-- Female-0.1196-1.030.20092.53 Number of workers-0.1461-2.31-- College-educated--0.24362.81 Income per household member---3.90 E-06-2.07 Household size--0.06653.12 Worker status0.37872.45-- VMT per household member-1.16 E-05-1.413.0 E-054.51 Age of home-0.076-2.44-0.049-2.27 Rooms in home0.08502.50-- Own home-0.1668-1.150.16661.63 Threshold 1-0.3473-1.24-1.429-8.51 Threshold 20.81792.93-0.1192-0.75 Threshold 31.2764.580.43122.69 Threshold 41.9967.251.2247.46 Log Likelihood at Convergence-1007.3-1050.0 Pseudo R 2 0.059900.02700