Apr 22, 2015
WATERSMART SOFTWARE
Bio
• Masters in Applied Econometrics
• Specialized in Regression Based Modeling (Specifically Logistic and Fixed Effects Regression Models)
• Learned about social comparison and demand response programs through years of evaluation work on energy efficiency programs.
• Apply those evaluation techniques to internal metrics at WaterSmart
• Design experiments and research projects to improve the efficacy of WaterSmart’s program
WATERSMART SOFTWARE
Home Water Report
Contents• WaterScore • Water use in GPD• Comparative consumption• Personalized
water-saving actions
WATERSMART SOFTWARE
California’s Response to the DroughtRegion• Who is responding to drought messaging, and what
influences their response?
• Production level data showing 5% decrease
• State Water Board Survey (July 2014)
• Details production level data for most utilities in CA
• Responses on drought related activities that each utility has taken
• Using Consumption level data for residential households we can isolate the response and match that up with the survey level data.
WATERSMART SOFTWARE
Regulatory Timeline
Jan 17, 2014Governor Brown Declares Drought State of Emergency
WATERSMART SOFTWARE
Jan 17, 2014Governor Brown Declares Drought State of Emergency
April 15, 2014Governor Brown Issues Executive Order to Redouble State Drought Actions
Regulatory Timeline
WATERSMART SOFTWARE
Regulatory Timeline
Jul 15, 2014State Water Board Adopts Emergency Conservation Regulations
“…mandatory restrictions on outdoor irrigation…”
Or seek approval for “an alternate plan that includes allocation-based rate structures”
WATERSMART SOFTWARE
WaterSmart California Dataset (July 2014)
Region• Clients that partner with WaterSmart transfer all of their residential billing data for the past 5 years.
• Usage data for ~ 5% of households within California
• Time period used for this analysis: Jan 2011-July 2014 (bimonthly readings)
• Apply conversion script to convert data from billing periods to calendar months and create full panel dataset
• Match with property database to acquire age of the home, lot-size, number of bedrooms, etc.
WATERSMART SOFTWARE
State Water Board Survey (July 2014)
Region
• Survey Responses from 276 out of 440 Water Suppliers in California
• Questions include:
• Total water production by month (2011-2013 average) and 2014.
• Drought response activities
• Forecasted shortages
• Aggregate analysis indicates 5% savings through May 2014 when compared to the average usage over the same period from 2011-2013.
WATERSMART SOFTWARE
• San Francisco (4,470 sq.mi.): The drainage into the Pacific Ocean from the Stemple Creek Basin boundary to and including the Pescadero Creek Basin.
• South Coast (11,100 sq.mi.): The drainage within the US that discharges into the Pacific Ocean from the Rincon Creek Basin boundary to the California-Baja California international boundary.
• Sacramento River (27,600 sq.mi.): The Sacramento River Basin and drainage into Goose Lake.
• San Joaquin (15,600 sq.mi): The San Joaquin River Basin.
WATERSMART SOFTWARE
Significant Variation between Production and Consumption Based Analyses (Production Data from State Water Board Survey and Residential Consumption Data from WaterSmart)
Hydrologic Region
Production Data
Consumption Data
San Francisco 8.7% 12.2%
South Coast -6.3% 0.2%
Sacramento River 13.0% 16.1%
San Joaquin River 26.4% 5.2%
Statewide(Sample) 5.0% 9.3%
San
Fran
cisc
o
Sout
h Coa
st
Sacr
amen
to R
iver
San
Joaq
uin
River
Stat
ewid
e0.0%
5.0%
10.0%
15.0%
20.0%
25.0%
Production DataConsumption Data
% Savings calculated using Feb – July Data
WATERSMART SOFTWARE
Response by Usage Groups(Evaluation of WaterInsight Program)
TR
EATM
EN
T E
FFEC
T (
%)
NO CHANGE
LowestUsers
HighestUsers
Analysis by Daniel Brent, Univ. of Washington, June 2012
WATERSMART SOFTWARE
Stat
ewid
e
Centra
l Coa
st
Sacr
amen
to R
iver
San
Fran
cisc
o
San
Joaq
uin
River
Sout
h Coa
st0.0%
5.0%
10.0%
15.0%
20.0%
25.0%
Top Water Users Slower to Reduce Consumption Top quartile versus bottom 3 quartiles
Hydrologic Region
Q1-Q3 Q4
San Francisco 15.9% 10.1%
South Coast 7.1% -8.1%
Sacramento River 20.4% 18.1%
San Joaquin River 7.4% 2.4%
Central Coast 20.5% 4.9%
Statewide(Sample) 13.8% 4.5%
Quartiles calculated using 2011-2013 average Gallons Per Day
WATERSMART SOFTWARE
San
Fran
cisc
o
Sout
h Coa
st
Sacr
amen
to R
iver
San
Joaq
uin
River
Stat
ewid
e0.0%
5.0%
10.0%
15.0%
20.0%
25.0%
Lot Size not a Significant Driver of Variation, Top quartile versus bottom 3 quartiles
Hydrologic Region
Q1-Q3 Q4
San Francisco 13.0% 13.2%
South Coast 4.2% -9.0%
Sacramento River 19.7% 17.8%
San Joaquin River 5.1% 4.8%
Statewide(Sample) 11.6% 7.9%
Quartiles calculated using property data for each household within each water supplier’s region
WATERSMART SOFTWARE
Positive Correlation Between Drought Response Actions and SavingsWater Supplier Action Correlation
with Savings
Drought Pricing 0.38
Mandatory Restrictions 0.46
Increased Enforcement and Monitoring 0.14
Drought Stage 0.71
Region
Drought Pric-ing
Mandatory Restrictions
Increased enforcement and monitor-
ing
Drought Stage
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
WATERSMART SOFTWARE
Enacted Mandatory Restrictions
No Mandatory Restric-tions
0.0%
5.0%
10.0%
15.0%
20.0%
25.0%
Water Suppliers that have Enacted Mandatory Restrictions have Higher Savings
Action Cumulative Savings since Jan
17th
Enacted Mandatory Restrictions 13.4%
No Mandatory Restrictions 2.8%
Region
Types of mandatory restrictions:• Restrictions on outdoor use
(e.g. limited irrigation schedules)
• Prohibitions on runoff into street and gutters
• Rationing• Customer Leak Repair
Requirement
WATERSMART SOFTWARE
Stage 0 Stage 1 Stage 20.0%
5.0%
10.0%
15.0%
20.0%
25.0%
Suppliers at Higher Drought Stage Levels Show Increased Savings
Utility Plan Stage Cumulative Savings since Jan
17th
Stage 0 -0.1%
Stage 1 8.9%
Stage 2 18.4%
Region
Drought stages are declared at the water supplier level in response to conditions in their region. Higher stages indicate a greater number of restrictions which can very based on each supplier’s drought plan.
WATERSMART SOFTWARE
More Conservation Actions Taken by a Utility Lead to Greater ReductionsDrought Response Actions
% Savings
0 1.8%
1 6.2%
2 9.6%
3 21.7%
4 18.7%
Region
0 1 2 3 40.0%
5.0%
10.0%
15.0%
20.0%
25.0%
Number of actions determined by aggregating utility actions: drought pricing, mandatory restrictions, increased enforcement and monitoring, and the drought stage.
WATERSMART SOFTWARE
PPIC Survey of Residential Households in California
Region
Data Set: 1,705 Adult Residents of California surveyed via phone (March 2014)
• “Would you say that the supply of water is a big problem, somewhat of a problem, or not much of a problem in your part of California?”
• Do you think that the water supply that is available for your part of California will be adequate or inadequate for what is needed ten years from now?
• The legislature is considering an approximately $11.1 billion dollar bond measure for the November 2014 ballot to pay for state water projects. If the election were being held today, would you vote yes or no on this state water bond?
• Would you say that you and your family have taken steps to reduce water use recently in response to the current drought?
WATERSMART SOFTWARE
Converting Survey Q’s to Binary Variables
Region• “Would you say that the supply of water is a big problem,
somewhat of a problem, or not much of a problem in your part of California?”
• If “big problem”, SupplyProblem = 1• Do you think that the water supply that is available for your
part of California will be adequate or inadequate for what is needed ten years from now?
• If “inadequate”, TenYrSupplyInadequate = 1• The legislature is considering an approximately $11.1 billion
dollar bond measure for the November 2014 ballot to pay for state water projects. If the election were being held today, would you vote yes or no on this state water bond?
• If “yes”, ProWaterBond = 1• Would you say that you and your family have taken steps to
reduce water use recently in response to the current drought?• If “yes”, Conserver = 1
• Merge with usage data.
WATERSMART SOFTWARE
Regression Results
Variable (1) (2) (3) (4) (5)
SupplyProblem 0.3548*(0.2185)
- - - 0.4274*(0.2219)
WaterBondImportant - -.09903(0.2150)
- - -0.2274(0.2211)
Conserver - - 0.1029(0.2157)
- 0.1873(0.2174)
ProEnvironmental - - - 0.1708(0.2151)
0.1397(0.2188)
N 882 872 881 872 863
R-Squared 0.01 0.01 0.01 0.01 0.01
• Regions with perceived local supply issues have responded stronger than those without.
• Other survey-based variables insignificant
Notes: (1) coefficient listed with SE in parenthesis. (2) * indicates coefficient is significant at the p=0.10 level.
WATERSMART SOFTWARE
Final Thoughts• Households respond to messaging at the local level, but
aren’t as responsive to state-wide messaging
• Instead of funding large state-wide campaigns, provide funding to local water suppliers for them to educate their customers.
• Mandatory restrictions work in pushing households to save water.
• Higher users and those with larger lots are not responding to messaging as strongly as others.
• Tend to be more affluent and not price sensitive
• Need to work on engagement with those households. That’s where social comparisons can be particularly effective.