Research and Evaluation of CAFE James M. Sallee The Harris School - - PowerPoint PPT Presentation

Research and Evaluation of CAFE

James M. Sallee

The Harris School University of Chicago

RFF December 17, 2013

Sallee (Harris) CAFE Review RFF December 17, 2013 1 / 10

• 1. Rational inattention
• Under some conditions, it may be rational for consumers to “ignore”

MPG when choosing a car (Sallee 2013)

• Elements of model:
• Consumer making discrete choice among vehicles
• Consumer has some easily determined (“free”) beliefs about value of

MPG for each car

• Suppose beliefs unbiased, but not perfectly accurate
• Consumer can pay additional cost (effort) to get perfect information, or

can make choice with imperfect, free information

• When will they be willing to pay effort cost?
• If they don’t pay effort cost, call this rational inattention to MPG

Sallee (Harris) CAFE Review RFF December 17, 2013 2 / 10

• 1. Rational inattention
• When will consumers choose inattention?
• Information is only valuable if it is pivotal

Sallee (Harris) CAFE Review RFF December 17, 2013 3 / 10

• 1. Rational inattention
• When will consumers choose inattention?
• Information is only valuable if it is pivotal
• Consider case of consumer choosing between A and B
• Assume A is first choice, given available information
• If consumer likes A a lot more than B, then MPG information unlikely

to be pivotal

• If variance in unknown fuel costs small, then MPG information unlikely

to be pivotal

Sallee (Harris) CAFE Review RFF December 17, 2013 3 / 10

• 1. Rational inattention
• When will consumers choose inattention?
• Information is only valuable if it is pivotal
• Consider case of consumer choosing between A and B
• Assume A is first choice, given available information
• If consumer likes A a lot more than B, then MPG information unlikely

to be pivotal

• If variance in unknown fuel costs small, then MPG information unlikely

to be pivotal

• Obviously, as costs of information acquisition rise, inattention more

likely

Sallee (Harris) CAFE Review RFF December 17, 2013 3 / 10

• 1. Rational inattention
• Fuel costs vary, but vehicle prices vary far more

Fuel Cost Price Price w/in VIN St Dev $3,100 $9,500 $2,400

Sallee (Harris) CAFE Review RFF December 17, 2013 4 / 10

• 1. Rational inattention
• Fuel costs vary, but vehicle prices vary far more

Fuel Cost Price Price w/in VIN St Dev $3,100 $9,500 $2,400

• Simulate choices using DCM of Langer (2013)
• Simulate data, estimate choice under full information, limited

information

• Calculate welfare loss from mistakes

Table : Simulation estimates from Sallee (2013)

Global mean Class Class, weight, cylinders Average welfare lost $522 $291 $89 (per vehicle purchased) % who change choice 19% 14% 7%

Sallee (Harris) CAFE Review RFF December 17, 2013 4 / 10

• 2. Information is costly
• Fuel economy labels do not resolve uncertainty entirely
• Need PDV
• EPA ratings imprecise, possibly biased

Sallee (Harris) CAFE Review RFF December 17, 2013 5 / 10

• 2. Information is costly/heterogeneity is large

5000 10000 15000 20000

Lifetime Fuel Cost ($)

15 20 25 30

Fuel Economy

Interquartile Range Median

Sallee (2013), borrows from Anderson, Kellogg and Sallee (2013)

• Heterogeneity in VMT, discount rate, gasoline price beliefs causes

idiosyncratic differences in valuation

• The market/gov’t cannot easily solve idiosyncratic uncertainty

Sallee (Harris) CAFE Review RFF December 17, 2013 6 / 10

• 2. Information is costly/heterogeneity is large

.05 .1 .15 .2 Density 15 20 25 30 Driver Fuel Economy (mpg)

On-road fuel economy by Driver

Langer and McRae (2013)

• Heterogeneity in on road MPG for identical car is vast
• Jacobsen, Knittel, Sallee and van Benthem (ongoing) studies policy

implications of heterogeneity

Sallee (Harris) CAFE Review RFF December 17, 2013 7 / 10

• 2. Rational inattention/information
• Consumers do not already have all the relevant information

(Turrentine and Kurani 2007)

• Policies that lower information costs unambiguously increase welfare
• But we don’t know how effective information provision can be
• More facts do not necessarily lower information costs
• Nudges that garner attention ambiguous

Sallee (Harris) CAFE Review RFF December 17, 2013 8 / 10

• 2. Rational inattention/information
• Consumers do not already have all the relevant information

(Turrentine and Kurani 2007)

• Policies that lower information costs unambiguously increase welfare
• But we don’t know how effective information provision can be
• More facts do not necessarily lower information costs
• Nudges that garner attention ambiguous
• Need evaluation of EPA labels [?]
• Field experiments will help (Allcott and Knittel)
• What should we learn from marketing experts?

Sallee (Harris) CAFE Review RFF December 17, 2013 8 / 10

• 3. Habit formation
• Consumer might have endogenous (habit formed) preferences over car

attributes based on prior experiences

• Creates possibility of multiple equilibria, may change welfare analysis

significantly

• E.g., could Americans be happy driving a European car fleet if they

were used to it?

• Anderson, Kellogg, Langer and Sallee (2013) shows that car buyers

influenced by the brand choice of their parents

Sallee (Harris) CAFE Review RFF December 17, 2013 9 / 10

• 4. Network/peer effects
• Policy MAY be warranted when there are network

externalities/consumer barriers to new technology adoption

• There are literatures on peer effects of technology, but limited for new

cars

• Heutel and Muehlegger (2013), Naranayan and Nair (2012) study

hybrids

• Bollinger and Gillingham (2012) study solar panels
• Development literature on agriculture
• Note that network effects can create hold up for automakers—do they

want to be leaders or followers in pioneering a technology?

• Answer determines whether policy needed, how it should be formed

Sallee (Harris) CAFE Review RFF December 17, 2013 10 / 10