Putting Behavioral Economics to Work:
Using Field Experiments to Manage
Residential Water/Energy Use
Michael Price
Department of Economics
Andrew Young School of Policy Studies
Georgia State University and NBER
1
Underlying Motivation – Broader Agenda
 Foster a deeper understanding of behaviors that generate public goods
(bads)
 Charitable contributions
 Private provision of environmental quality (conservation and compensation)
 Identify influences that drive such actions
 Highlight what models best predict behavior and outline directions for new
theories
 Provide guidance for policymakers and practitioners
2
Underlying Motivation – Broader Agenda
 Examine mechanisms – both monetary and non-monetary – to
address underlying externality
 Charitable lotteries or characteristics of a solicitor
 Social comparisons or subsidies
 Provide an apples-to-apples comparison of commonly employed
policies
 Uncover channels through which mechanisms influence behavior to
inform theory and design of new policies
 Avoid policies or actions that are ineffective or promote unintended
consequences
3
Experiment #1 –
Compliance and Temporal Patterns of Use
 Restrictions that limit number of days a week
household can water lawns
 Enforcement of regulations is problematic
 Infrequent water patrols
 Nominal fines for repeated violations in same calendar year
4
Experiment #1 –
Compliance and Temporal Patterns of Use
 Daily monitoring project of 4,800 residential water
consumers over eight week period in summer 2007
 Readings are taken overnight from households with smart meter
technology
 Households face mixture of flat-rate and increasing block pricing
 Households randomly assigned to either a control group or
one of three treatment conditions
 Schedule reminder
 Drought letter with pro-social appeal
 Monitoring letter highlighting unusual patterns of use in the area
5
Experiment #1 –
Compliance and Temporal Patterns of Use
 Treatment letters mailed during fourth week of project
 Identification of treatment effects based on difference-indifferences approach
 Compare change in use after intervention across treatment
and control group
 Subset of households are monitored following summer
to examine persistence
6
Experiment #1 –
Compliance and Temporal Patterns of Use
 Estimate an approximate 23% likelihood of watering on
unassigned day in pre-intervention period
 Post-intervention rates of compliance
 No difference in rates of compliance amongst control group
 Schedule reminder generates 9.5% reduction in rate of noncompliance
 Normative appeal generates a 6.5% reduction in rate of noncompliance
 Monitoring letter generates a 15.2% reduction in rate of noncompliance
7
Experiment #1 –
Compliance and Temporal Patterns of Use
 Average daily use on unassigned days
 No discernible difference in use after intervention amongst
control group
 Significant reduction in use after intervention – 6.4 to 11.9
percent – amongst treatment groups
 Average daily use on assigned days
 Significant increase in use for households assigned the
schedule and monitoring letters
 Significant reduction in use for households receiving
normative appeal
8
Experiment #2 –
Promoting Conservation Efforts
 Cobb County Water System
 Second largest user of public water supplies in state
 Distributes treated surface water to approximately 170,000
customers
 Partnered with CCWS to implement norm-based
conservation campaign in summer 2007 (drought)
 Information campaigns highlighting how and why to
conserve water
 Apples-to-apples comparison of appeal to civic duty and
social comparisons
9
Experiment #2 –
Promoting Conservation Efforts
 Households randomized into four treatment cells and
use tracked for three year period
 Control group
 Group that received technical advice letter on how to
conserve
 Group that received technical advice letter and norm-based
appeal to conserve
 Group that received technical advice letter and norm-based
appeal that included social comparison
10
Experiment #2 –
Promoting Conservation Efforts
 Messages focusing on why to conserve more effective
than those stating how to conserve
 No significant reduction in use amongst HHs in technical
advice treatment
 HHs in social comparison treatment consume approximately
4.8 percent less than those in the control group
 Impacts are heterogeneous and more pronounced
amongst highest user groups
11
Experiment #2 –
Promoting Conservation Efforts
 Examine persistence of effects by looking at use
during summer 2008 and summer 2009
 Social comparisons have lasting impact on water
consumption
 Consume 2.6 percent less than counterparts in control during
summer 2008
 Consume 1.3 percent less than counterparts in control during
summer 2009
 Unable to detect long-run effect for normative appeal
12
Experiment #3 –
Adoption of Green Technologies
 Two motivations for adoption of green technologies
 Altruism
 Social pressures
 Is adoption decision supply or demand driven?
 Welfare implications and “costs” to utility differ across
motives
 Altruism and welfare enhancing whereas social pressures can
be welfare reducing
13
Experiment #3 –
Adoption of Green Technologies
 Consider two specific policy instruments
 Price reductions ($5 or $1)
 Use of a descriptive norm (social comparison – local or national)
 Households in suburbs of Chicago approached by a paid
solicitor and offered opportunity to purchase CFLs
 Allow subset of the households ability to sort into or out of
interaction
 Households are warned in advance of visit but vary ease with
which they can avoid solicitor
14
Experiment #3 –
Adoption of Green Technologies
 Conditioned on answering the door, HHs who received
warning are more likely to purchase CFLs
 Importance of altruism for subset of individuals
 Prices and social norms influence purchase decision
 Prices influence both the decision to purchase and number of
packages purchased
 Normative appeals only influence the decision to purchase
15
Experiment #3 –
Adoption of Green Technologies
 Increase in price from $1 to $5 generates an approximate
62.5% reduction in purchases
 Suggests that demand for CFLs is inelastic
 Comparison of prices and descriptive norms
 Inclusion of “high” social norm statement is equivalent to
approximate 70% reduction in prices ($5 to $1.50)
 Inclusion of a “low” social norm statement is equivalent to an
approximate 30% reduction in prices ($5 to $3.50)
16
Experiment #4 –
Salience, Norms and Energy Demand
17
Experiment #4 –
Salience, Norms and Energy Demand
18
Experiment #4 –
Salience, Norms and Energy Demand
 Information intervention to HHs with historical
average consumption between 100 and 125 kWh
 Random assignment to 3 treatment (16K HHs in each)
 Make the 111 kWh price notch salient
 Include a social comparison with reference consumption at
notch
 Cross the two messages
19
Experiment #4 –
Social Comparisons and Norms…
20
Experiment #4 –
Price Salience….
21
Major Lessons Learned
 Both neo-classical and behavioral motives are important
drivers of behavior
 Norms matter
 Salience (or lack thereof) reduces the sensitivity to prices
 Prices matter….
 Important complementarities between motives and policies
22
Major Lessons Learned
 Impacts of behavioral interventions are more
pronounced in short-run and tend to wane over time
 Out of sight, out of mind
 Boy who cried wolf
 Significant heterogeneity in the effects of such
strategies across observable dimensions
23
Some Take Away Thoughts….
 No one size fits all policy to promote demand
management
 Drivers of behavior are heterogeneous
 Impacts differ across households and over time
 Partnerships between public utilities and academics
have proven highly successful
 Informed policy and helped achieve desired outcomes
 Furthered our understanding of household behavior and
factors that drive demand
24