Welfare Effects of Home Automation Technology with Dynamic Pricing - - PowerPoint PPT Presentation

▶
welfare effects of home automation technology with
SMART_READER_LITE
LIVE PREVIEW

Welfare Effects of Home Automation Technology with Dynamic Pricing - - PowerPoint PPT Presentation

Jun 14, 2023 358 likes •573 views

Welfare Effects of Home Automation Technology with Dynamic Pricing Bryan Bollinger Wes Hartmann Duke University Stanford University EI Workshop, May 2016 Welfare Effects of Home Automation Technology with Dynamic Pricing 1 / 20 Motivation

slide-1
SLIDE 1

Welfare Effects of Home Automation Technology with Dynamic Pricing Bryan Bollinger Wes Hartmann

Duke University Stanford University

EI Workshop, May 2016

Welfare Effects of Home Automation Technology with Dynamic Pricing 1 / 20

slide-2
SLIDE 2

Motivation

How can we better enable consumers to respond to changing market conditions? Electricity Demand Application:

◮ Prices are regulated ◮ Real-time pricing seems ideal ◮ Frequency of effective price variation depends on timeliness of response ◮ Policymaker perspective: Can we reduce reliance on more polluting supply

alternatives?

◮ Utility perspective: Can we delay additional capacity investments?

Can we combine rich historical data with field experiments to better target treatments?

Welfare Effects of Home Automation Technology with Dynamic Pricing 2 / 20

slide-3
SLIDE 3

What We Do

Non-parametrically estimate average consumer welfare impacts of information and home automation technologies

◮ Use control group demand to condition on unobserved demand shocks

Estimate the firm surplus from providing all consumers with each technology Estimate household-level treatment effects for each technology x price treatment combination

◮ Match households based on distributions of pre-treatment consumption as a

function of state

Estimate the firm surplus with optimal targeting

Welfare Effects of Home Automation Technology with Dynamic Pricing 3 / 20

slide-4
SLIDE 4

Hardware Options for Dynamic Price & Response

In-Home Display (IHD): provides information in convenient location Programmable Communicating Thermostat (PCT): automates response

Welfare Effects of Home Automation Technology with Dynamic Pricing 4 / 20

slide-5
SLIDE 5

Price Treatments

Time of Use with Critical Pricing:

◮ Off-peak price of 4.2c per kWh Weekends, off-peak times on weekdays and holidays ◮ On-Peak price of 23c per kWh On-peak times (2p-7p) on weekdays ◮ Critical price of 46c per kWh Rare price overcall for 2 to 8 hours at any time during the year

Variable Peak Pricing:

◮ Off-peak price of 4.5c per kWh Weekends and off-peak times on weekdays ◮ On-Peak prices average 13.6c per kWH:

  • Price / kWh

4.5 11.3 23 46 Freq in Days 50 37 23 10

Communicated by 5pm the day before ◮ Critical price of 46c per kWh Requires only 2 hours notice

Control Pricing:

◮ 8.4c per kWh regardless of day or time ◮ Increases to X after Y cumulative kWh in the month Welfare Effects of Home Automation Technology with Dynamic Pricing 5 / 20

slide-6
SLIDE 6

Experiment & Treatment Assignments

Treatments CON TOU VPP Total Control 294 294 Portal 310 332 642 IHD 224 228 452 PCT 214 220 434 All three 184 172 356 Total 294 932 952 2,178 Includes some households without AC who were eventually dropped

Welfare Effects of Home Automation Technology with Dynamic Pricing 6 / 20

slide-7
SLIDE 7

Average Hourly Usage by Treatment

1 1.5 2 2.5 3 3.5 4 Usage (KWH) 0:00 14:00 19:00 24:00 control portal only ihd and portal pct and portal all 3

Weekdays (w/AC)

1 1.5 2 2.5 3 3.5 4 Usage (KWH) 0:00 14:00 19:00 24:00 control portal only ihd and portal pct and portal all 3

Weekends, (w/AC)

Welfare Effects of Home Automation Technology with Dynamic Pricing 7 / 20

slide-8
SLIDE 8

Average Effects (w/AC)

Table: Average Treatment Effect Regression Results

Critical Peak Off-Peak Treatments TOU VPP TOU VPP TOU VPP Portal

  • 0.437
  • 0.281
  • 0.324
  • 0.146
  • 0.113

0.031 (0.137) (0.136) (0.123) (0.123) (0.082) (0.086) IHD

  • 0.671
  • 0.414
  • 0.491
  • 0.274
  • 0.138
  • 0.018

(0.141) (0.137) (0.127) (0.126) (0.090) (0.087) PCT

  • 1.190
  • 1.219
  • 0.942
  • 0.681

0.027 0.118 (0.141) (0.141) (0.127) (0.130) (0.087) (0.094) All 3

  • 1.082
  • 1.231
  • 0.830
  • 0.643

0.080 0.080 (0.134) (0.149) (0.131) (0.138) (0.097) (0.102) N=2,141 N=2,178 N=2,178 Standard errors in parentheses Greater reduction in TOU likely attributable to higher average price

Welfare Effects of Home Automation Technology with Dynamic Pricing 8 / 20

slide-9
SLIDE 9

All 3 vs. IHD Demand Curves

0.0 0.1 0.2 0.3 0.4 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

Price kW PCT/IHD, 10th perc. IHD, 10th perc. PCT/IHD, 50th perc. IHD, 50th perc. PCT/IHD, 90th perc. IHD, 90th perc.

Figure: All 3 vs. IHD Demand: 10th, 50th and 90th Percentiles of Control Demand

Welfare Effects of Home Automation Technology with Dynamic Pricing 9 / 20

slide-10
SLIDE 10

All 3 vs. IHD Expenditure Differences

0.0 0.1 0.2 0.3 0.4 0.5 1 2 3 4 5 Price kW 0.0 0.1 0.2 0.3 0.4 0.5 1 2 3 4 5 Price kW 0.0 0.1 0.2 0.3 0.4 0.5 1 2 3 4 5 Price kW

Welfare Effects of Home Automation Technology with Dynamic Pricing 10 / 20

slide-11
SLIDE 11

All 3 vs. IHD Welfare Differences

0.0 0.1 0.2 0.3 0.4 0.5 1 2 3 4 5 Price kW 0.0 0.1 0.2 0.3 0.4 0.5 1 2 3 4 5 Price kW 0.0 0.1 0.2 0.3 0.4 0.5 1 2 3 4 5 Price kW

Welfare Effects of Home Automation Technology with Dynamic Pricing 11 / 20

slide-12
SLIDE 12

All 3 vs. IHD Welfare Differences

Aggregating across control states, consumers receive a welfare gain of about $28 over the summer, or $280 NPV.

◮ Breaking the linearity of demand between prices gives bounds of $213 to $486

But, in addition to consumer welfare, the firm benefits from reducing demand during periods of critical peak demand.

◮ The costs of supplying a PCT total $250 ◮ PCT needs to decrease critical period usage by -250/700 = -0.357 kW ◮ PCT (with portal) demand reductions are, on average, nearly a full kWh ◮ But can the utility do better? Welfare Effects of Home Automation Technology with Dynamic Pricing 12 / 20

slide-13
SLIDE 13

Household Treatment Effects

We match HHs based on their distribution of usage as a function of state variables (time of day, temperature, and recent electricity consumption.). We use a weighted regression for each HH to eatimate that HH’s treatment effect.

Welfare Effects of Home Automation Technology with Dynamic Pricing 13 / 20

slide-14
SLIDE 14

Effects of Portal by TOU and VPP

difference in control and portal coef.

  • 3
  • 2
  • 1

1 2 3 cdf 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 critical peak

  • ff peak

difference in control and portal coef.

  • 3
  • 2
  • 1

1 2 3 cdf 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 critical peak

  • ff peak

Welfare Effects of Home Automation Technology with Dynamic Pricing 14 / 20

slide-15
SLIDE 15

Effects of IHD by TOU and VPP

difference in control and ihd coef.

  • 3
  • 2
  • 1

1 2 3 cdf 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 critical peak

  • ff peak

difference in control and ihd coef.

  • 3
  • 2
  • 1

1 2 3 cdf 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 critical peak

  • ff peak

Welfare Effects of Home Automation Technology with Dynamic Pricing 15 / 20

slide-16
SLIDE 16

Effects of PCT by TOU and VPP

difference in control and pct coef.

  • 3
  • 2
  • 1

1 2 3 cdf 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 critical peak

  • ff peak

difference in control and pct coef.

  • 3
  • 2
  • 1

1 2 3 cdf 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 critical peak

  • ff peak

Welfare Effects of Home Automation Technology with Dynamic Pricing 16 / 20

slide-17
SLIDE 17

Effects of All Three by TOU and VPP

difference in control and all 3 coef.

  • 3
  • 2
  • 1

1 2 3 cdf 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 critical peak

  • ff peak

difference in control and all three coef.

  • 3
  • 2
  • 1

1 2 3 cdf 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 critical peak

  • ff peak

Welfare Effects of Home Automation Technology with Dynamic Pricing 17 / 20

slide-18
SLIDE 18

For Which Households Will a PCT Cut 0.357 kWh

portal

  • 3
  • 2
  • 1

1 2 3 pct

  • 3
  • 2.5
  • 2
  • 1.5
  • 1
  • 0.5

0.5 1 1.5 pct vs portal, tou peak portal

  • 3
  • 2
  • 1

1 2 3 pct

  • 3
  • 2.5
  • 2
  • 1.5
  • 1
  • 0.5

0.5 1 1.5 pct vs portal, vpp peak

Welfare Effects of Home Automation Technology with Dynamic Pricing 18 / 20

slide-19
SLIDE 19

CDF of critical treatment effect

difference pct and portal coef.

  • 3
  • 2
  • 1

1 2 3 cdf 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 TOU VPP

Welfare Effects of Home Automation Technology with Dynamic Pricing 19 / 20

slide-20
SLIDE 20

In Conclusion

Information can improve demand response to variable pricing,

◮ But response technology is much more important (for electricity)

Home automation increases consumer welfare but largest gains are to the firm Significant heterogeneity in response can be described by historical usage data

◮ Demographics have little explanatory power

Targeted distribution of technologies can avoid installations where the effects

  • f the technology do not justify the cost.

◮ Increases the surplus per installation by 30%

Integrating extensive historical behavior into field experiments could yield similarly efficient targeted treatments

◮ Medical treatments condition on limited observable characteristics, but digital

daily health data could reveal unobserved differences in treatment effects

Welfare Effects of Home Automation Technology with Dynamic Pricing 20 / 20