Rational Inattention and Energy Efficiency
James M. Sallee
The Harris School
University of Chicago
June 21, 2011
Sallee (Harris)
Inattention
June 21, 2011
1 / 23
• Notches are ubiquitous in energy policy
A Kink
A Notch
Total
Tax
Total
Tax
Energy Efficiency
Energy Efficiency
• Ex: Efficiency standards, Energy Star, LEED, Feebates, Hybrid Credit,
Gas Guzzler Tax
Sallee (Harris)
Inattention
June 21, 2011
2 / 23
• Notches are ubiquitous in energy policy
A Kink
A Notch
Total
Tax
Total
Tax
Energy Efficiency
Energy Efficiency
• Ex: Efficiency standards, Energy Star, LEED, Feebates, Hybrid Credit,
Gas Guzzler Tax
• Notches in externality-correcting policies are inefficient (Sallee &
Slemrod 2010)
- Pigouvian tax efficient because everyone has same marginal incentive
- Notches provide uneven incentives – may not get low hanging fruit
Sallee (Harris)
Inattention
June 21, 2011
2 / 23
• Notches are ubiquitous in energy policy
A Kink
A Notch
Total
Tax
Total
Tax
Energy Efficiency
Energy Efficiency
• Ex: Efficiency standards, Energy Star, LEED, Feebates, Hybrid Credit,
Gas Guzzler Tax
• Notches in externality-correcting policies are inefficient (Sallee &
Slemrod 2010)
- Pigouvian tax efficient because everyone has same marginal incentive
- Notches provide uneven incentives – may not get low hanging fruit
• Are there benefits to notches that might outweigh efficiency
costs?
Sallee (Harris)
Inattention
June 21, 2011
2 / 23
What are plausible benefits of notches?
• Administrative ease
- “Lawyers prefer tables to formulas”
- “Staffers can’t put logarithms into legalese”
• Salience
- Notches create first-order welfare changes
- Notches simplify (coarsen) consumer information
- Fine-grained information might slip through net of consumer attention,
whereas coarse information is snagged
Sallee (Harris)
Inattention
June 21, 2011
3 / 23
What are plausible benefits of notches?
• Administrative ease
- “Lawyers prefer tables to formulas”
- “Staffers can’t put logarithms into legalese”
• Salience
- Notches create first-order welfare changes
- Notches simplify (coarsen) consumer information
- Fine-grained information might slip through net of consumer attention,
whereas coarse information is snagged
Why might coarse information be more effective than fine-grained
information?
• Assume some cognitive cost of processing information
• Could be rational to ignore fine-grained information if expected gain
small relative to cognitive cost
• Is it rational for consumers to ignore energy efficiency when
purchasing durables?
- Could justify notches
- Could explain energy gap
Sallee (Harris)
Inattention
June 21, 2011
3 / 23
Goal of this paper is to explore whether or not rational inattention is
plausible.
Sallee (Harris)
Inattention
June 21, 2011
4 / 23
Goal of this paper is to explore whether or not rational inattention is
plausible.
Research summary
1
2
Energy efficiency notches create distortions
Model: inattention more likely rational as...
• Variance in energy costs smaller
• Variance in other attributes larger
• Cost of learning larger
3
Rational inattention is plausible
• Learning about energy efficiency is valuable
• But, variation in other attributes quite large
• Learning is difficult
Sallee (Harris)
Inattention
June 21, 2011
4 / 23
Producers respond to notches
Tax
4000
40
0
0
2000
20
Weighted Count
6000
60
8000
Vehicles that paid the Gas Guzzler Tax
12
14
16
18
20
22
Fuel Economy (mpg)
Source: Sallee & Slemrod 2010
Sallee (Harris)
Inattention
June 21, 2011
5 / 23
Producers respond to notches
.1
Tax<22.5
0
.05
Density
.15
Cars subject to Gas Guzzler Tax
15
20
25
30
35
40
Fuel Economy (mpg)
Source: Sallee & Slemrod 2010
Sallee (Harris)
Inattention
June 21, 2011
6 / 23
Producers respond to notches
0
.05
Density
.1
.15
LEED certification (NC 2.2)
20
30
40
50
60
70
Points
Sallee (Harris)
Inattention
June 21, 2011
7 / 23
Producers respond to notches
0
.05
Density
.1
.15
LEED certification (NC 2.1)
20
30
40
50
60
70
Points
Sallee (Harris)
Inattention
June 21, 2011
8 / 23
Producers respond to notches
0
.05
Density
.1
.15
LEED certification (CI 2.0)
20
30
40
50
60
70
Points
Sallee (Harris)
Inattention
June 21, 2011
9 / 23
Producers respond to notches
0
.05
Density
.1
.15
LEED certification (CS 2.0)
20
30
40
50
60
70
Points
Sallee (Harris)
Inattention
June 21, 2011
10 / 23
Producers respond to notches
.6
Energy Star (refrigerators)
0
.2
Density
.4
ES>20%
0
10
20
30
Percent Above Federal Standard
Sallee (Harris)
Inattention
June 21, 2011
11 / 23
Producers respond to notches
.6
Energy Star (freezers)
0
.2
Density
.4
ES>10%
0
10
20
30
Percent Above Federal Standard
Sallee (Harris)
Inattention
June 21, 2011
12 / 23
Producers respond to notches
Energy Star (washers)
.04
0
.02
Density
.06
ES<=324
200
250
300
350
400
kWh/year
Sallee (Harris)
Inattention
June 21, 2011
13 / 23
Heuristic model of rational inattention
• Consumer makes discrete choice between models
• Uncertain of exact lifetime fuel cost
• Can learn exact costs after incurring search cost
• Does consumer pay information cost, or just choose without resolving
uncertainty?
Sallee (Harris)
Inattention
June 21, 2011
14 / 23
• Two good example:
- Risk neutral consumer i
- Models j and k
- Difference in utility not including energy cost surprise is
θi = Vj + εij − Vk − εik
- Energy cost surprise is c ∼ N (0, σc ) for both j and k (iid)
- Search cost is s
• Consumer i will search iff:
√
(1 − Φ(θi / 2σc )) ·
"
√
#
√
φ(θi / 2σc )
√
− θi > s.
2σc
1 − Φ(θi / 2σc )
• Search more likely as...
- utility difference between models θi smaller
- search cost s gets smaller
- variance in energy costs σc bigger
Sallee (Harris)
Inattention
June 21, 2011
15 / 23
Search decision for different θi , holding fixed σc and s
Choose k
with search
Choose j
with search
Choose k
without search
Choose j
without search
A -Θ* B
Θ*
• Consumers very far from indifferent θi > |θ ∗ | will not search; choose
without knowing true costs
• Consumers who like j and k about the same will search
• Inattention need not induce underprovision
- If −cj + ck = B, then all consumers who would change purchase search;
no mistakes are made and demand is a function of true energy cost
Sallee (Harris)
Inattention
June 21, 2011
16 / 23
Automobile variation
Median and standard deviation in price and fuel cost by vehicle type
40,000
35,000
30,000
25,000
20,000
15,000
10,000
5,000
0
Compact Car Midsize Car Luxury Car Sports Car
Price
Sallee (Harris)
SD Price
Fuel Cost
Inattention
SUV
Pickup
Van
SD Fuel Cost
June 21, 2011
17 / 23
Automobile variation
Standard deviation in price within VIN and fuel cost across VIN
3,500
3,000
2,500
2,000
1,500
1,000
500
0
Compact Car Midsize Car Luxury Car
Sports Car
SD Price w/in VIN
SUV
Pickup
Van
SD Fuel Cost
• Within VIN variation exceeds within class fuel cost variation
• Model year cycle: save ≈ $150/month; one standard deviation change
in timing ≈ $750
• How hard is it to calculate fuel costs?
• Don’t labels solve this problem?
Sallee (Harris)
Inattention
June 21, 2011
19 / 23
• How hard is it to calculate fuel costs?
• Don’t labels solve this problem?
• Information incomplete and imprecise
30
25
20
10
15
5 Cycle Rating
35
40
Variation in highway fuel economy between test methods
10
15
20
25
30
35
40
Original Rating
• EPA changed test procedure in 2008
• Mean change in lifetime fuel costs is $1,700
• Standard deviation in change in cost is $600
Sallee (Harris)
Inattention
June 21, 2011
19 / 23
Appliance variation
Median and standard deviation in price and energy cost by appliance
2,500
2,000
1,500
1,000
500
Sallee (Harris)
s
Re
fri
Ra
ge
ng
ra
es
to
r(
Re
to
fri
p
,a
ge
ut
ra
o)
to
Re
r
fri
(si
de
ge
,a
ra
to
ut
r(
o)
bo
t
t
om
Fr
ee
,a
ze
ut
r(
o)
up
rig
ht
,a
ut
o)
Fr
ee
ze
r(
ch
es
t)
wa
sh
es
Price
O
ve
n
t)
fro
n
er
(
er
wa
sh
ot
h
Cl
Cl
ot
h
es
Di
sh
wa
sh
e
rs
(to
p)
0
SD Price
Fuel Cost
Inattention
SD Fuel Cost
June 21, 2011
20 / 23
• What about appliance labels?
Based on standard U.S. Government tests
Refrigerator-Freezer
With Automatic Defrost
With Side-Mounted Freezer
With Through-the-Door-Ice Service
U
XYZ Corporation
Model ABC-W
Capacity: 23 Cubic Feet
Compare the Energy Use of this Refrigerator
with Others Before You Buy.
This Model Uses
600kWh/year
�
Energy use (kWh/year) range of all similar models
Uses Least
Energy
539
Uses Most
Energy
698
kWh/year (kilowatt-hours per year) is a measure of energy (electricity) use.
Your utility company uses it to compute your bill. Only models with 22.5 and 24.4
cubic feet and the above features are used in this scale.
Refrigerators using more energy cost more to operate.
This model's estimated yearly operating cost is:
$54
Based on a 2005 U.S. Government national average cost of 9.064 per kWh for electricity. Your
actual operating cost will vary depending on your local utility rates and your use of the product.
Important: Removal of this label before consumer purchase violates the Federal Trade Commission’s Appliance Labeling Rule (16 C.F.R. Part 305).
Sallee (Harris)
Inattention
June 21, 2011
21 / 23
Lifetime cost without discounting and with 5% discounting
1800
1600
1400
1200
1000
800
600
400
200
0
Dishwashers
Clothes
Washers
Refrigerators Clothes Dryers
CAN "Second Tag"
Ranges
Freezers
Corrected (r=5%)
• Natural Resources Canada provides “second price tag” = lifetime cost
of energy
• But, they calculate this without discounting
• Mistake creates large error in cost estimates
• Mistakes much larger than true standard deviations in fuel costs
Summary
• Markets respond to notches ⇒ implies inefficiency
• Rational inattention plausible
- Costs of information significant
- Price variation large
Extensions
• Smaller appliances better candidates for inattention
• Discrete choice estimation to quantify distance, match model
• Second choice data to enable direct counting of mistakes
• Field experiments
Sallee (Harris)
Inattention
June 21, 2011
23 / 23