Using Feedback as a Tool for Household Energy Conservation: An Experimental Approach
Kannika Thampanishvong
Policy Dialogue “Transition to Green Economy for Southeast Asia”
6-7 March 2015
Background
• In Thailand, little emphasis has been placed on energy conservation by consumers in the residential sector.
• There is a good potential for electricity saving• Behavioral changes in the use of residential energy
could contribute towards overall energy consumption.• This research project puts emphasis on tools that
encourage household energy conservation.
Why Electricity Conservation Nudge?
• Residential electricity market tends to suffer from asymmetric information
• Study the effect of information provision on behavioral changes.
• Highlight the importance of information communicated to households and the way in which information is presented and framed.
• What type of information ? Energy saving “hints” Telling people how their energy consumption compares
to that of their “neighbours”
Why Electricity Conservation Nudge?
• Residential electricity market tends to suffer from asymmetric information
• Study the effect of information provision on behavioral changes.
• Highlight the importance of information communicated to households and the way in which information is presented and framed.
• What type of information ? Energy saving “hints” Telling people how their energy consumption compares
to that of their “neighbours”
What do other studies tell us?
• Ferraro and Price (2011 found that social comparison messages are most effective among high-users but the effectiveness of such messages wanes over time.
• Ayres et al. (2009) investigated the impact of provision of peer-comparison feedback to customers and found that mailed energy feedback has effect on energy consumption.
Key Research Questions
1. Does the feedback on electricity consumption successfully motivate households in the treatment groups to reduce energy consumption?
2. Which form of feedback is the most effective? Which treatment group experiences the largest reduction in electricity consumption?
Data Authorization Process
❶ ❷ ❸ ❹
Who are the Participants?
Minburi
Nongjok
Ladkrabang
Sapansung
Kannayao
Klongsamwa
161 households from 10 housing estates
How Participants are Assigned into Groups?
161 HHs
G1: Control Group, no feedback
G2: Treatment Group1,
peer comparison feedback
G3: Treatment Group 2, energy saving tips
G4: Treatment Group3,
peer comparison and energy saving tips
Randomly
assigned
Information Kits for Group 2
Information Kits for Group 3
Information Kits for Group 4
Variable NameTreatment 1
(n= 31)
Treatment 2
(n= 36)
Treatment 3
(n= 36)
Control
(n= 35)
Average Household Size (persons) 3.3 3.3 3.1 3.2
Average Proportion of Elderly Household Members Aged 60 and above in the Household (percent)
23 14 16 8
Average Proportion of Young Household Members Aged 15 and below in the Household (percent)
9 10 14 18
Average Proportion of Student in the Household (percent)
14 19 31 39
Average Household Income (THB) 61,936 57,196 75,579 71,152
Average Proportion of Households that Own Property (percent)
94 94 94 97
Average Proportion of Household Members who Stay at Home during Daytime (percent)
37 29 37 29
Average Length of Residence in Current Home (years)
9.5 8.6 10.6 9.8
Average House Size (squared wa) 33.8 32.2 32.6 37.3
Average Proportion of Household Member with at least Secondary Education (percent)
77 81 82 77
Average Proportion of Female Member in the Household (percent)
57 50 52 56
Mean Comparisons of Pre-Treatment Variables
Pre-treatment Electricity Consumption
Source: Author’s own computation using data from MEA
Impacts of Households’ Characteristics on Pre-treatment Consumption
Households’ Characteristic Variables [1] [2]Household size 73.3537***
(14.8797)
Dummy (whether some members in the household stay at home during daytime)
79.1410***
(30.1856)
130.2590***
(38.6551)
Dummy (whether the house is owned by household) 0.7852
(52.8593)
-10.7158
(46.7181)
Length of residence in current home 3.9063
(4.4709)
2.2910
(4.9951)
House size 2.8313**
(1.1227)
3.1962***
(1.2247)
Dummy (whether there are elderly members aged 60 and above in the household)
12.2098
(47.9286)
Dummy (whether there are young members aged 15 and below in the household)
68.9946**
(37.2236)
Constant -54.1696
(67.3181)
133.3841
(52.3726)
Which Group Perform Better?
Source: Author’s own computation using data from MEA
Difference-in-Difference Results
Group Mean Electricity Consumption during Pre-Treatment Period
[1]
Mean Electricity Consumption during
Post-Treatment Period
[2]
[1] – [2] Difference-in-Difference
(kWh)
Control Group 402.58 400.71 1.87
Treatment 1 346.48 324.27 22.20 20.34
Treatment 2 415.53 404.10 11.43 9.56
Treatment 3 480.92 451.95 28.97 27.10
Fixed Effect Model Estimations,
[1] [2]Treatment Dummies
d_Treatment1
d_Treatment2
d_Treatment3
-20.6128
(13.8932)
-7.4584
(17.4705)
-25.0497*
(12.8066)
-21.1481
(13.7610)
-7.5574
(17.4691)
-24.8587*
(12.6382)
Climate Variable
Temperature 18.9252
(12.0741)Number of Observations
R-squared within
3025
0.3986
3025
0.3995
Treatment Group 3 that received both peer comparison feedback and electricity saving hints make significant reductions in electricity consumption.
Percentage Reduction in Electricity Consumption
Pre-treatment consumption(unit: kWh)
Reduction in Consumption(unit: kWh)
Percentage Reduction in Electricity Consumption
(unit: Percentage)
Treatment 1 346.48 -21.0948 -6.09%
Treatment 2 415.53 -7.3213 -1.76%
Treatment 3 480.92 -24.8587 -5.17%*
Average Changes in Electricity Consumption for Treatment 3
Policy Messages
• Electricity conservation nudge in the form of peer comparison and electricity saving tips works in term of influencing reduction in electricity consumption.
• MEA and PEA: cost-effectiveness of sending electricity consumption feedback together with electricity bill
• Possible extensions: Further testing for different forms of “hints” with different
treatments containing more specific electricity saving tips Testing whether effects wane over time