how do consumers respond to water and electricity pricing? · city limits laguna woods san diego...

How Do Consumers Respond to Water and Electricity Pricing?

Evidence from Recent Empirical Studies in Economics

Koichiro ItoSIEPR Postdoctoral Fellow

Stanford [email protected]

Assistant Professor, Boston University (from July 2013)

mailto:[email protected]


Two Common Assumptions in Demand Estimation and Forecasting

1) Consumers fully understand their price schedule

2) Economic theory tells us that consumers respond to marginal price

2Example: Nonlinear Residential Water Pricing in Irvine Ranch Water District in CA

Figure 3: Five-Tier Increasing Block Price Schedule in IRWD in August 2002

0

1

2

3

4

5

6

Mar

gina

l pric

e ($

) per

CC

F

0 40 100 150 200 250 300Monthly consumption relative to baseline allocation (%)

Notes: This figure shows the five-tier increasing block price schedule in IRWD in August 2002. IRWD

allocates“baseline allocation”to a customer, and the customer’s marginal price depends on consumption

relative to the baseline allocation. The marginal price equals the first tier rate up to 40% of the baseline,

the second tier rate up to 100%, the third tier rate up to 150%, the fourth tier rate up to 200%, and

the fifth tier rate over 200% of the baseline.

26

Potential Problems about the Two Assumptions

3

Consumers Are Not Well Informed about their Water or Electricity Prices

4


Easy to see what gas price you are paying

5


Easy to see what gas price you are paying

6

Hard to see what water price you are paying

Typical Utility Bills: Difficult to Understand

7

Example: An Electricity Bill in California

5

Typical electricity bills are unclear and difficultto get the information about prices

MOTIVATION 3)

Ɣ Electricity bills – Hard to get the information about prices

Typical Utility Bills: Difficult to Understand

8

Example: A Water Bill in California

Given this environment,Do Consumers Respond to Correct

Marginal Price?

9

I Examine How Consumers Actually Respond to Water and Electricity Prices

Introduction Research Design Estimation Welfare Conclusion

2) Changes in price schedules can be used to estimate demand

4lnxit = ↵+ �4lnput(xit) + "it

Previous studies use simulated instruments (policy-induced price changes):

4lnpPIut (xit) = lnput(xit0)� lnput0(xit0)

Typically, the first stage is very strongAn identification assumption: a parallel trend between A and B

Consumption

Price

Household A Household B

35 / 82

Partnered with Water Utility (Irvine Ranch Water District) in CA







Consumption

Price


35 / 82

Partnered with Electricity Utilities in CA







Consumption

Price


35 / 82

Household-level monthly billing data for about 10 years







Consumption

Price


35 / 82

Electricity consumers: 40,749 households







Consumption

Price


35 / 82

Water consumers: 64,601 households







Consumption

Price


35 / 82

Quasi-experimental research design







Consumption

Price


35 / 82

Use policy changes as natural experiments







Consumption

Price


35 / 82

Exploit spatial discontinuities to create “treatment” and “control” groups

10

In Orange County CA, Households in the Same City Have Different Power Companies

11

Nonlinear electricity pricing Randomized Field Experiments on Dynamic Electricity Pricing

I exploit a nearly ideal research environment in electricity markets in California

Edison (Southern California Edison) provides electricity for the north side

LagunaBeach

AlisoViejo

LagunaNiguel

Laguna Hills

MissionViejo

Cote deCoza

LasFlores

RanchoSanta

Margarita

Border of Electric Utility Service Areas

City Limits

Laguna Woods

San Diego (San Diego Gas & Electric) provides electricity for the south side3 / 8

They Experience Very Different Pricing

12

Nonlinear electricity pricing Randomized Field Experiments on Dynamic Electricity Pricing

Households experience substantially different nonlinear pricing

Edison and San Diego: Cents per kWh in 2002

San Diego

Edison

10

15

20

25

Monthly Consumption

4 / 8

Average Price = Total Bill Payment / Total Usage

Similarly, Exploit Policy Changes in Irvine Ranch Water District (IRWD)

13

Figure 2: Service Areas of Irvine Ranch Water District (IRWD)

Notes: This figure shows service ares in Irvine Ranch Water District (IRWD). The District serves the

City of Irvine and portions of the Cities of Costa Mesa, Lake Forest, Newport Beach, Tustin, Santa

Ana, Orange and unincorporated Orange County. Santa Ana Heights service area was consolidated in

1997 and Lake Forest service are was consolidated in 2001. The original map is provided by IRWD.

25

Residential Water Pricing Changed from Flat Pricing to Nonlinear Pricing

14

Figure 3: Five-Tier Increasing Block Price Schedule in IRWD in August 2002

0

1

2

3

4

5

6

Mar

gina

l pric

e ($

) per

CC

F

0 40 100 150 200 250 300Monthly consumption relative to baseline allocation (%)

Notes: This figure shows the five-tier increasing block price schedule in IRWD in August 2002. IRWD

allocates“baseline allocation”to a customer, and the customer’s marginal price depends on consumption

relative to the baseline allocation. The marginal price equals the first tier rate up to 40% of the baseline,

the second tier rate up to 100%, the third tier rate up to 150%, the fourth tier rate up to 200%, and

the fifth tier rate over 200% of the baseline.

26

(1) Before Policy Change (Flat)

(2) After Policy Change (Nonlinear)

What Do I Find?

15

Findings: Both Water and Electricity Consumers Respond to Average Price (not Marginal Price)

16


Research question: How do consumers respond to nonlinear price schedules?

1 Standard economic theory predicts:

Consumers respond to Marginal Price

2 Laboratory experiments find:

Many individuals respond to Average Price = ( Total payment / Quantity )

Demand Curve

MP

AP

10

15

20

25

Ma

rgin

al P

rice

(ce

nts

pe

r kW

h)

Monthly Consumption

) Standard theory and laboratory evidence provide different predictions

6 / 81

Why Do We Care about the Findings?

1) Responding average price weaken the incentive for conservation

2) Forecasts based on marginal price might be biased

17


Research question: How do consumers respond to nonlinear price schedules?

1 Standard economic theory predicts:

Consumers respond to Marginal Price

2 Laboratory experiments find:

Many individuals respond to Average Price = ( Total payment / Quantity )

Demand Curve

MP

AP

10

15

20

25

Ma

rgin

al P

rice

(ce

nts

pe

r kW

h)

Monthly Consumption

) Standard theory and laboratory evidence provide different predictions

6 / 81

So, What Can We Do?

18

The Key is:Providing Better Price Information

19

In My Other Research, I Provide “In-home display” for Electricity Consumers

20

Figure 2: Information on In-Home Display

&),"��01

% �� ,$��-�� #�+*!��'��!��.��0(/��

Notes: This figure shows an example screenshot of the in-home display that are installed for bothcontrol and treatment consumers in the experiment.

16

Using the “In-home display”, Consumers Can See Real-Time Information about Price and Usage

21

Figure 1: Variable Critical Peak Pricing (VCPP)

15

50

75

100

150

0

25

50

75

100

125

150

Price

pe

r kW

h (

JPY

)

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Time of the day

Notes: This figure shows the price schedule of the variable critical peak pricing in the experiment.The marginal price of electricity is 15 yen for every hour in non-event days. When a CPP event dayis announced, consumers are informed about their critical peak price one-day ahead, which is either50, 75, 100, or 150 yen.

15

Randomized Field Experiment

Households

22

Treatment Group

Control Group

Random Assignment

When Consumers Have Clear Price Information, They DO Respond to their Price Incentives Correctly

23

Figure 3: Mean Consumption of Treatment and Control Groups by Peak-time Price

!2

!1.5

!1

!2

!1.5

!1

!2

!1.5

!1

10 15 20

10 15 20

15 50

75 100

150

Log o

f E

lect

rici

ty U

se p

er

30!

min

ute

inte

rval

Time of the dayGraphs by tp

Notes: This figure shows the mean of log consumption per 30-minute interval for the control (� ) andtreatment (⇧) consumers. I use data from all CPP event days to calculate the means. The verticallines show the critical peak time period, 1 pm to 5 pm.

17

Dynamic Pricing Group (Orange) and Control Group (Green)

Similar Findings from Other Studies: Providing Better Information is the Key







Consumption

Price


35 / 82

Teaching Income Tax Code:







Consumption

Price


35 / 82

Chetty and Saez (2013)







Consumption

Price


35 / 82

Other Evidence from Electricity Pricing







Consumption

Price


35 / 82

Wolak (2011), Jessoe and Rapson (2012), Kahn and Wolak (2013)

24

Summary: What Can We Learn from Recent Economic Studies?

1) Not-clear price information --> Consumers don’t get right price signals







Consumption

Price


35 / 82

Evidence: both water and electricity consumers respond to average price







Consumption

Price


35 / 82

Wrong price signals --> weaken incentives for conservation

2) Providing better price information is the key







Consumption

Price


35 / 82

Evidence from electricity pricing and income taxation







Consumption

Price


35 / 82

Consumers respond to price signals correctly when they receive clear info

3) Discussion







Consumption

Price


35 / 82

How can we improve the clarity of water price information for consumers?

25

Thank you for your attention

Koichiro ItoSIEPR Postdoctoral Fellow

Stanford [email protected]



how do consumers respond to water and electricity pricing? · city limits laguna woods san diego...

Documents