FTC Staff Comment Before the National Highway Traffic Safety Administration Concerning Environmental Impact Statement for the Corporate Average Fuel Economy (CAF) Program

V890096
COMMISSION AUTHORIZED
BEFORE THE
NA nONAL HIGHWAY TRAFFIC SAFETY ADMINISTRATION
Washington, D.C.
)
)
latent to Prepare aa Enviroamental 	)
Impact Statement Cor the Corporate 	)
A verage Fuel Economy Program 	)
)
------------)
..9 CFR Part 531
[Docket No. FR-89-19••1]
COMMENTS OF THE STAFF OF THE BUREAU OF ECONOMICS
OF THE FEDERAL TRADE COMMISSION-
Washington, D.C. 	20580
November 13, 1989
• 	These 	comments 	are 	the 	views 	of 	the 	staff 	of 	the 	Bureau 	of
Economics of the Federal Trade Commission. 	They are not necessarily the
- 	views of the Commission or of any individual Commissioner.

I. latrod.ctloa .ad S ••••ry
On 	September 	12, 	1989, 	the 	National 	Highway 	Traffic 	Safety
Administration (NHTSA) issued a Notice of Intent asking for comments about
the 	potential 	environmental 	impacts 	of 	Corporate 	Average 	Fuel 	Economy
(CAFE) standards on automobiles.1 	Comments are requested to aid NHTSA in
the preparation of an Environmental Impact Statement (EIS) for 	the CAFE
program. 	A major environmental effect of the CAFE program comes from
changes 	in gasoline consumption due to the enforcement of the standards.
In response 	to NHTSA's request. the staff of the Bureau of Economics of
the 	Federal 	Trade 	Commission 	("the 	staff") 	submits 	the 	foJlowing
conceptual 	and 	quantitative 	analysis 	of 	the 	consequences of 	higher 	CAFE
standards fOT 	gasoline consumption. 2 	For 	standards set for 	years close at
hand, 	these 	comments 	present 	both 	conceptual 	analysis 	and 	quantitative
estimates 	of 	possible 	changes 	in 	gasoline 	consumption. 	While 	we 	do 	not
present 	quantitative 	estimates 	of 	the 	effects 	of 	standards 	set 	for 	years
farther 	in 	the 	futvre. 	we 	present 	conceptual 	analysis 	concerning 	these
effects.
Economic theory suggests that a binding CAFE standard can have several
effects on 	gasoline 	consumption. 	Firms may choose 	to meet 	the standard
either 	by 	engaging 	in 	technological 	innovation 	or 	by 	selling 	relatively
1 54 	Federal 	Register 	37702 	(September 	12. 	1989). 	There 	is 	a
similar 	program 	in 	place 	for 	light 	trucks 	which 	we 	""ill 	not· discuss 	in
- 	these comments.
2 These 	comments 	are 	the 	views 	of 	the 	staff 	of 	the' Bureau 	of
Economics of the Federal Trade Commission. 	They are not necessarily the
views 	of 	the 	Commission 	or 	of 	any 	individual 	Commissioner. 	Inquiries
regarding these comments should be directed to staff economist Andrew N.
Kleit, 	Bureau 	of 	Economics, 	Federal 	Trade 	Commission. 	Washinlton. 	D.C..
20580, (202) 326-3481.
1

, .
more 	(uel-efficient 	and 	(ewer 	(uel-intensive 	vehiclcs. 	Through 	thcse
means, 	CAFE 	standards 	can 	push 	consumcrs 	into 	more 	tuel-erricicDt
vehicles 	than 	they 	otherwisc 	might 	usc. 	This, 	by 	itsclf, 	would 	servc 	to
decreasc 	,&soline 	consumption. 	There 	are, 	however, 	other 	effects 	of 	a
binding CAFE standard that may lerve to affect fuel 	consumption. 	These
comments 	discuss 	four 	luch 	effects 	and 	factor 	them 	into 	estimates 	of
,&soline lavings provided 	by 	various CAFE standards. 	First, 	by 	changing
the 	price of new cars, CAFE standards can 	change 	the 	scrappage 	rate 	of
used 	cars. 	Second. 	by 	increasing 	the 	fuel-efficicncy 	of 	newer 	vehicles,
CAFE standards can lower thc cost of driving additional miles, which may
increase 	the 	number 	of 	miles 	travelled. 	Third, 	by 	altering 	the 	relative
prices of automobiles. CAFE standards can change the number of vehicles on
the 	road. 	Finally, 	by 	constraining 	the 	sales 	of 	less 	fuel-efficient
vehicles by companies that face a binding CAFE constraint. CAFE standards
can 	indirectly 	decreasc 	the 	fuel-efficiency 	of 	the 	vehicles 	sold 	by 	firms
that do not face a binding CAFE constraint.s
The current NHTSA model may mis-estimate the consumption savings from
higher 	CAFE 	standards, 	if, 	as 	appears 	to 	be 	the 	casc, 	it 	does 	not 	take
account 	of 	the 	factors 	noted 	abovc. 	Wc 	will 	prcscnt 	estimates 	that
indicate 	that 	consideration 	of 	such 	factors 	can 	significantly 	change 	the
expected gasoline savings from the CAFE program. 	Thcrefore, we urge that
NHTSA take these factors into account in any EIS that attcmpts to measure
the conservation effcct of CAFE standards.
I 	In 	the 	body 	of 	the 	comments 	we 	refcr 	to 	thcse 	errects 	as 	the
-scrappagc 	errect,- 	the 	-mileage 	effect,- 	the 	-markct 	elasticity 	effect."
and the -market substitution effect,- respectively.
2

II. Experleace .f tlae Staff .f tlae Federa' Trade Co •• laaloD
The Federal Trade Commission (FTC) is an independent relulatory aaency
rcsponsible 	(or 	(osterina 	competition 	and 	safeauarding 	the 	iDterests 	of
consumers.· 	Upon 	request 	by 	(ederal, Itate, and 	local 	aovernment 	bodies,
the 	Itaff 	of 	the 	FTC 	(requently 	analyze 	regulatory 	proposals 	to 	identify
provisions 	that 	may 	impair 	competition 	or 	increase 	costs 	to 	consumers
witbout 	offering 	compensatory 	benefits. 	Tbe 	FTC 	also 	bas 	a 	statutory
responsibility 	to 	review 	complaints 	tbat 	the 	CAFE 	standards 	harm
competi tion.5
The FTC staff has accumulated significant experience in examining the
effects of CAFE standards over tbe past few years. 	In 1986, the FTC staff
lubmitted comments when NHTSA was considering modifying CAFE standards for
Model Years (MY) 1987 and 1988.s 	In 1988 the staff of the FTC's Bureau of
Economics submitted comments when NHTSA was considering modifying CAFE
• 	IS U.S.C. section 4J et seq.
5 Tbe Secretary of the Department of Transportation 	may 	modify 	or
waive 	CAFE 	civil 	penalties 	under 	certain 	limited 	circumstances, 	one 	of
which is certification by the FTC that waiver or modification is necessary
to 	prevent 	barm 	to 	competition. 	IS 	U.s.C. 	I 	2008 	(b)(3)(C). 	The 	statute
provides that a manufacturer may petition the FTC for a certification that
modification 	of 	tbe 	civil 	penalty 	-is 	necessary 	to 	prevent 	a 	substantial
lesseninl of competition in that segment of the automobile industry subject
to the standard with respect to which such penalty was assessed.- JS V.S.C I
2008 (b)(4).
I 	Comments of the Bureaus of Competition, Consumer Protection, and
Economics of tbe Federal Trade Commission, Passenger Automobile A.,erage
I=wl 	Economy 	Standards 	Model 	Year 	1987-88, 	National 	Higbway 	Traffic
Safety Administration, Docket No. FE-IS-OJ, March 26, 	1986. 	(Hereinafter
cited as FTC Staff Comments (1916).)
3

".' l
atandards (or MYs 1919 and 1990.' 	In addition, members of the FTC staff
have conducted a series of studies of the effects of the CAFE law,'
m. Descriptio. of the CAFE Prolram
The CAFE prOlram, as enacted in 1975, calls (or all auto manufacturers
producing 	more than 	10,000 automobiles per 	year 	to satisfy 	the 	mandated
CAFE levels. 	The program establishes average MPG standards that apply to
a 	manufacturer's 	entire 	Cleet, 	rather 	than 	to 	the 	(uel-efficiency 	of
indivjdual models. 	CAFE levels were scheduled to rise from 	18.0 MPG in
1978 	to 	27.S 	MPG 	in 	Model 	Year 	(MY) 	1985 	and 	all 	subsequent 	years.
Failure 	to satisfy 	the 	MPG standard 	can 	result in 	the imposition 	of 	civil
penalties. 8
The 	CAFE 	legislation 	divided 	each 	manufacturer's 	fleet 	into 	two
distinct groups - 	foreign and domestic. 	Average MPG for a manufacturer's
, Comments of the Staff of the 	Bureau of Economics of the Federal
Trade ColDllliJsion to the 	National Highway Traffic Safety Administration,
"'Panenger Automobile Average Fuel Economy Standards for Model Years 1989
and 	1990,- September 	IS, 	1988. 	(Hereinafter cited as FTC Staff Comments
(1988).)
, See A. Kleit, eyhe Economics of Automobile Fuel Economy Standards,w
Ph.D. 	dissertation, 	Yale 	University, 	1987. 	Also see 	Kleit, eyhe 	Impact 	of
Auto 	Fuel 	Standards" 	FTC 	Working 	Paper 	No. 	160, 	February 	1988, 	and
-Enforcing Time Inconsistent Government Regulations,w FTC Working Paper
No. 	161, March 	1988; R. Rogers, eyhe Short-Run Impact of Changes in the
Corporate 	Average 	Fuel 	Economy 	Standards" 	Bureau 	of 	Economics 	Staff
Manuscript, March 1986.
8 The 	Cine 	(or 	a 	Cirm 	out 	of 	compliance 	is 	equal 	to 	S50 	times 	the
Dumber 	of 	cars 	made 	by 	the 	producer 	in 	a 	liven 	model 	year 	times 	the
amount 	(in 	miles 	per 	lallon) 	that 	the 	producer 	is 	out 	o( 	compliance
(public Law .6:1S-2008). 	So Car, however, no domestic manufacturers, and
only a few Imall European firms, have actually paid fines.

domestic 	cars 	aDd 	Coreian 	can 10 	are 	each 	calculated 	aeparately.ll 	The
provision was deuaned to iDduce domestic firms to produce hiah-MPG cars
iD the U.s.12
Although the CAFE ItaDdards are let by Itatute,1I NHTSA is empowered
to adjust 	throuah 	rulemakiDgs 	the 	CAFE Itandard 	for 	MY 	1985 and 	any
subsequent year to the -maximum feasible average fuel economy level.- 	In
determining 	this 	maximum 	feasible 	level. 	NHTSA 	is 	required 	to 	consider
four 	ractors: 	(I) 	technological 	f easi bili ty, 	(2) 	economic
practicability. 	(3) 	the 	effect 	of 	other 	federal 	motor 	vehicle 	standards 	on
fuel economy. and (4) the need of the Dation to conserve energy.l. 	NHTSA
acted to lower the required MPG standard from 27.5 MPG to 26.0 MPG for MYs
J986 through 1988 and from 27.5 lOG to 26.5 MPG for MY 1989. The standard
for MY 1990 remains at the original statutory level of 27.5 MPG.
During 	the 	time 	that 	CAFE 	standards 	have 	been 	in 	force, 	average
automobile 	fuel 	efficiency 	has 	risen 	significantly. 	The 	limited 	available
quantitative 	evidence 	indicates, 	however. 	that 	a 	lignificant 	reason 	for
this 	rise 	has 	been 	the 	increase 	in 	aasoline 	prices 	rather 	than 	the
imposition of standards. 	Researcb by Crandall and Graham15 indicates that
10 A foreign 	car is defined as one having less than 75% of its value
added produced in the United States.
11 53 Federal Register 33080 (August 29. 1988).
12 NHTSA Annual Report on Fuel Economy, 1982, 9.
11 Title V of the Motor Vehicle Information and Cost Savings Act, IS
U.s.c. 2001 el seq.
14 53 Federal Register 33081 (August 29. 1988).
II Robert W. Crandall and John D. Graham, "The Effect of Fuel EcoDomy
Standards on Automobile Safety.- Jour1tQl of Law and Economics 32:2 (April,
1989). 97-) 18.
s

only a 	amall 	part 	of the 	increase 	in 	fuel-efficiency 	can 	be attributed 	to
CAFE ltandards Corcing firms to change the mix of the CIeet of cars lold.
Crandall and Graham also Cind no lignificant technological innovations in
automobile 	Cuel 	efficiency 	introduced 	as 	a 	result 	of 	CAFE 	standards
between 1978 and 1986.
IV. 	Economic Effects of CAFE Standards That May Affect
Gasoline ConlenatioD
Domestic 	automotive 	lasoline 	consumption 	is 	determined 	by 	three
factors: 	(I) 	the 	average 	number 	of lallons 	of lasoline 	required 	to drive
a 	mile, 	(2) 	the 	average 	number 	of 	miles 	driven 	per 	vehicle, 	and 	(3) 	the
total 	number 	of 	vehicles 	owned 	by 	American 	consumers. 	By 	esta blishing
standards Cor newly produced cars, CAFE regulations directly address only
the 	first 	of 	these 	factors. 	If 	the 	latter 	two 	factors 	do 	not 	change
subsequent 	to 	the 	imposition 	of 	CAFE 	standards, 	and 	if 	the 	rate 	of
replacement 	of 	existing 	large 	cars 	is 	not 	reduced 	in 	response 	to 	the
standards, 	then 	the 	program 	may 	fulfill 	its 	objective 	of 	reducing 	total
fuel 	consumption. 	It may 	be 	unlikely, 	however, 	that 	these 	factors 	would
remain 	constant. 	Changes 	in 	these 	factors 	can 	be expected 	to 	influence
actual gasoline savings.
When CAFE standards become binding 18 on a particular firm, that firm
Ie A ~inding· CAFE standard 	is 	one 	that 	changes 	the average 	fuel-
eCficiency 	of 	the 	automobiles 	produced 	compared 	to 	the 	averale 	Cuel-
eCficiency 	that 	wouid 	have 	resulted 	if DO CAFE standard existed. 	Thus,
for example, if a firm's fleet averages 3S MPG aDd a CAFE staDdard of 27.S
MPG were established, the standard would not be bindinl on that firm. 	If
the 	firm's 	CIeet averages 	2S MPG, the standard 	would 	be binding on that
firm.
6

.as 	two 	methods 	by 	which 	it 	can 	raise 	its 	CAFE 	rating. 	First, 	it 	can
cn.ale 	in 	-technololY 	forcing· 	by chanlinl 	the 	technololY 	in 	iu vehicles
so 	that 	they 	become 	more 	fuel-efficient. 	Second, 	in 	both 	the ahort and
the 	long 	run, 	the 	firm 	can 	engage 	in 	what 	may 	best 	be 	termed 	as -mix-
shifting.·17
Technology 	forcing 	is 	a 	long-run 	phenomenon 	because 	it 	generally
requires 	enlineering 	modifications 	that 	could 	take 	several 	years 	to
implement. 	Thus, such actions are more likely to occur 	if firms 	are told
of 	the 	relevant 	CAFE 	standard 	several 	years 	in 	advance. 	Further, 	for
technology-forcing 	to 	occur, 	firms 	must 	have 	reason 	to 	believe 	that 	the
standard 	is 	likely 	to 	be 	upheld 	once 	the 	date 	for 	its 	enforcement 	is
reached. 	Otherwise, 	they 	may 	not 	have 	sufficient 	incentives 	to 	increase
the 	fuel-efficiency 	of 	their 	vehicles 	to 	the 	mandated 	standard. 	Such
innovation 	may, 	while 	lowering 	the 	per-mile 	costs 	of 	driving, 	generate
substantial production costs which may be passed on to consumers,
If 	technology 	forcing 	is 	used 	to 	meet 	a 	particular 	CAFE 	standard,
aasoline 	savings 	will 	be 	influenced 	by 	the 	·scrappage 	effect: 	While
teehJtological 	'PrOl'f~~ m1ly 	increase 	new 	car 	average 	fuel-efficiency 	over
time, 	the 	improvements 	may 	lead 	to 	an 	increase 	in 	the 	price 	of 	new
vehicles that may be greater than the savings to consumers in reduced fuel
operating 	costs. 	To 	the 	extent 	that 	this 	is 	true, 	the 	scrappage 	rates 	of
used 	cars, 	which 	are 	likely 	to 	be 	less 	fuel-efficient 	than 	new 	cars, 	will
17 In the long run a firm has the choice of meeting a standard either
by 	-technology-forcing· or 	-mix-shifting,· 	It can 	be 	expected to use 	both
in a manner such that the marginal cost to the firm of using either method
is equal. 	See Kleit (J9Ub), IllmI. note 8.
7

be 	reduced 	(and 	the 	price 	of 	used 	cars 	will 	rise).18 	Of 	course. 	the
relative 	Cuel efficiency of new cars versus used cars depends on the rate
of 	technololical 	innovation. 	which 	in 	turn 	depends 	on 	such 	factors 	as
psoline prices and the level of relevant CAFE standards.l~
Gasoline ovinls lenerated by technology (orcing may also be influenced
by the -mileage effect.- 	A binding CAFE standard may induce consumers to
buy 	more 	(uel-efficient 	vehicles 	than 	they 	otherwise 	would. 	Since 	the
lasoline 	cost 	of 	driving 	additional 	miles 	for 	these 	consumers 	is 	reduced,
they likely will be induced to drive more miles than they otherwise would.
It 	has 	been 	estimated 	that 	a 	10 	percent 	increase 	in 	fuel-efficiency 	of
particular 	vehicles 	leads 	to an 	increase 	in 	miles 	driven 	of 	approximately
3 	percent,1O 	Thus, 	a 	10 	percent 	increase 	in 	fuel-efficiency 	can 	be
expected 	to 	decrease 	fuel 	consumption 	slightly 	less 	than 	7 	percent 	for
those consumers who are 	induced 	to 	buy 	more 	fuel-efficient 	vehicles than
they otherwise would.
Firms also may 	meet 	binding standards set 	in 	years close at hand 	and
those 	set 	to 	take 	effect 	several 	years 	in 	the 	future 	by 	mix-shifting.
1. This 	phenomenon 	has 	been 	observed 	in 	other 	regulatory 	contexts.
For example, it has been shown that increasing the stringency of pollution
controls 	can 	actually 	increase 	pollution 	(in 	net 	present 	value 	terms) 	by
delaying 	the 	replacement 	of 	older, 	more 	pollution-intensive 	automobiles.
See 	Howard 	K. 	Gruenspecht, 	-Differentiated 	Regulation: 	A 	Theory 	with
Applications 	to 	Automobile 	Emission 	Controls: 	Yale 	University 	Ph.D.
dissertation, 	1982. and 	Gruenspecht. -Differential 	Regulation: 	The 	Case 	of
Auto Emissions Standards,- American Economic Review 72:3 (1982), 328-331.
Ie If 	technological 	progress 	does 	lenerally 	increase 	fuel-efficiency
over 	time, 	it 	implies 	that 	for 	CAFE standards to 	be 	remain 	binding they
will have to rise over time.
10 See Roger D. Blair, David L. Kaserman. and Richard C. Tepel. --rhe
Impact of 	Improved 	Gasoline Mileage on Gasoline 	Consumption,- Economic
/lIfIuiry (1914) 209-217.
1

Mix-shifting 	means 	that 	a 	manufacturer 	will 	raise 	the 	price, 	and 	reduce
the 	quantity 	IOld 	(relative 	to 	the 	prices 	and 	quantities 	otherwise
dictated by market forces). of cars having low-MPG ratings. and reduce the
price, 	and 	raise 	the 	quantity 	lold, 	of 	cars 	offering 	high-MPG 	ratings.
ID 	effect, 	mix-shifting 	implies 	that 	the 	automaker 	has 	established 	an
internal 	system 	of 	quasi-taxes 	and 	subsidies 	for 	its 	autos.21 	The 	change
in 	price 	of 	each 	type 	of 	car 	depends 	primarily 	on 	the 	relation 	of 	that
car's 	fuel-efficiency 	to 	the 	CAFE 	standard 	(the 	less 	fuel 	efficient 	a
particular model, the greater the internal tax).
Like 	technology 	forcing, 	mix-shifting 	will 	contribute 	to 	the 	-scrappage
effect- 	and 	the 	-mileage 	effect.- 	Mix-shifting 	may 	cause 	two 	additional
effects. 	First. mix-shifting can 	generate what we refer to as the -market
elasticity 	effect.- 	CAFE 	standards 	implicitly 	act 	as 	a 	tax 	on 	new 	large
cars and as a subsidy on new small cars." 	Depending on demand and supply
conditions. 	the 	increase 	in 	the 	number 	of 	small 	cars 	produced 	(and
subsequently 	purchased 	by 	consumers) 	could 	more 	than 	offset 	the 	decline
in 	large 	cars 	produced 	(and 	subsequently 	purchased 	by 	consumers).!! 	In
general. this would tend to diminish the extent to which fuel 	consumption
would 	decline 	if 	large 	car 	production 	fell 	by 	as 	much 	as 	small 	car
production 	rose. 	On 	the 	other 	hand, 	if 	the 	increase 	in 	small 	car
J1 By 	artificially 	decreasing 	large 	car 	production, 	and 	increasing
Imall 	car 	production, 	a 	higher 	CAFE 	standard 	is 	conceptually 	similar 	to
placing 	a 	tax 	on 	large 	cars, 	and 	a 	subsidy 	on 	small 	cars. 	See 	John 	E.
Kwoka Jr., -Yhe Limits of Market Oriented Regulatory Techniques: The Case
of 	Automotive 	Fuel 	Economy 	Standards,- 	Quarterly 	Journal 	01 	Economics
(1913). 69S-704. and K.leit (1987) and (1988).
12 See the discussions in K.leit (1988) and FTC Staff Comments (1988).
II See K. wolea. Illm1 at 702.
9

production were 1D0re than offset by the decrease in large car production,
CAFE 	Itandards could 	certainly decrease 	the Dumber 	of 	Dew cars on 	the
road. 	This would tend to increase the decline in aasoline consumption as
• result of CAFE Itandards.
Second. mix-shiftinl also may aenerate what we have termed the -market
substitution effect.- 	In leneral. lome firms (those with fleet 	MPGs above
the relevant level) may not be constrained by a 	particular CAFE standard.
Manufacturers unconstrained by CAFE Itandards will have incentives to sell
laraer. 	less 	fuel-efficient 	vehicles. 	This 	is 	because 	their 	rivals 	will
DOt be able to lDeet the demand for these types of automobiles and thus the
prices of these automobiles will rise. 	Thus, while CAFE standards can be
expected to raise the averale fuel 	efficiency of those firms 	bound 	by the
standards. 	CAFE 	standards 	may 	also 	lower 	the 	fuel 	efficiency 	of 	those
firms Dot bound by the standards. depending upon those firms' competitive
strategies aDd distance from the CAFE standard.
For 	example. 	currently 	General 	Motors 	and 	Ford 	appear 	to 	be
constrained 	in larle car production 	by 	CAFE standards 	while 	Chrysler 	is
DOt. 	Chrysler would thus appear to have a comparative advantage over GM
and 	Ford 	in 	the 	production 	of 	new 	fuel-efficient 	vehicles. 	Consequently,
if 	newly 	binding 	CAFE 	Itandards 	cause 	the 	prices 	for 	large 	new 	cars
produced 	by OM and 	Ford 	(e.I.. 	Cadillacs and 	Lincolns) to 	rise, 	Chrysler
might 	be 	able 	to 	increase 	its 	profits 	by 	producing 	more 	of 	its 	large
models (e.g .• the 	New Yorker) without jeopardizing its compliance with the
standard. 	Thus. 	Chrysler's 	average 	fleet 	MPG 	could 	decline 	if 	CAFE
standards required a rise in the fleet MPGs of GM and Ford. 	(This example
10

also applies to several Coreiln manufacturers who are not bound by CAFE
Itandards.)
Mix-ahiftinl 	may also 	have important 	impacts on 	the scrappase effect.
CAFE 	Itandards 	may 	act 	to 	raise 	the 	price 	of 	new 	larle 	can. 	which
implies 	that 	consumers 	will 	hold 	used 	large 	cars 	longer. 	and 	therefore
will 	scrap 	them 	later. 	This 	implies 	that 	in 	the 	near 	term 	used 	large
cars. which can 	be expected 	to 	be less fuel-efficient 	than 	new 	luge cars,
wiII stay on the road longer and consume more S8Soline. 	Mix-shifting also
implies. 	however. 	that 	the 	price 	of 	new 	small 	cars 	relative 	to 	the 	price
of luge cars may be reduced, which implies that owners of used small OT
large cars may replace them with new small cars. which can be expected to
be 	morc 	fuel-efficient 	First-time 	purchasers 	accordingly 	may 	be 	more
inclined to buy a small car than they otherwise would be.
There 	are 	factors 	other 	than 	the 	four 	effects 	enumerated 	above 	that
may 	also 	affect 	the 	estimates 	of 	changes 	in 	gasoline 	consumption 	due 	to
CAFE standards. 	For instance, CAFE standards on automobiles. which raise
the 	price 	of 	relatively 	fuel-inefficient 	automobiles. 	may 	increase 	the
demand for light-duty trucks. 	If more light-duty trucks were purchased,
aasoline 	consumption 	would 	litel'y 	increase 	because 	light-duty 	trucks
presently have Jess stringent CAFE standards than automobiles (around 20.5
MPG as opposed to 27.5 MPG).'· 	Unfortunately. we are currently unaware of
any 	empirical 	estimates 	that 	would 	aid 	in 	determining 	the 	potential
magnitude of such switcbina. 	Nor can we Sfl' how long the differential in
CAFE 	standards between 	light 	trucks 	and 	automobiles 	will 	continue. 	We
14 Light truck MPG standards have been set in the range 20.0 to 20.5
MPG for MYs 1916 through 1991. 54 Federal Register 37703 (September 12. 1989).
11

therefore 	do 	Dot 	iDclude 	this 	effect 	iD 	aDy 	of 	the 	estimates 	that 	we
present in Section V.
The 	facton 	discussed 	iD 	this 	section 	may 	have 	importaDt 	impacts 	on
aDy estimates of lasoline consumption livings due to CAFE ItaDdards. 	It
appears that these factors are Dot currently taken 	account of in NHTSA's
model of lasoline savings due to CAFE Itandards. t5
V. SlmulatloD Results of CAFE StaDdards OD GasoliDe CODsenatloD
In 	this section. we lenerate quantitative estimates of the magnitude of
the 	Cour 	effects 	discussed 	in 	Section 	IV 	(the 	scrappage 	effect. 	the
mileage 	effect. 	the 	market 	elasticity 	effect. 	and 	the 	market 	substitution
effect). 	The 	model 	we 	presented 	in 	FTC 	Staff 	Comments 	(1988) 	uses
various 	assumptioDs 	to 	lenerate 	estimates 	of 	the 	aasoline 	savings 	that
would have occurred had NHTSA chosen CAFE standards above 26.6 MPG for MY
1989, 	(26.6 MPG is the highest level at which a CAFE standard would have
been 	Donbinding for 	major producers in MY 	1989.) 	A description of this
model is contained in the technical appendix to these comments,
We 	Dote 	that 	this 	model 	is 	designed 	to 	calculate 	the 	·short 	run"
effects of higher CAFE standards. 2e 	It was devised to address whether to
25 Our knowledge of NHTSA's method for estimating lasoline savings
due 	to 	CAFE 	standards 	comes 	from 	·Final 	Reaulatory 	Impact 	Analysis:
Average Fuel Economy Standard for Model Year 1989 Passenger Automobiles,"
Plans and Policy Office of Relulatory Analysis. National Hiahway Traffic
Safety Administration, September 28. 1989 (·RIA").
Ie By ·short run" we mean that the model is designed to measure the
effects 	of 	NHTSA's 	lowering 	or 	raising 	the 	CAFE 	standard 	under 	its
administrative 	discretion 	immediately 	prior 	to 	a 	particular 	model 	year 	or
ror 	a 	year 	close 	at 	hand. 	Under 	such 	conditions, 	weUlume 	that
manufacturers 	do 	Dot 	hive 	the 	opportunity 	to 	enille 	iD 	"echnology
rorcing", 	It therefore must meet the hilher CAFE standard solely by -mix-
shifting".
12

cbanle the CAFE standard for a model year close at band, a question NHTSA
bas 	addressed 	several 	times since 	1985. 	It 	is alSumed 	in 	tbe model 	that
firms reach the bilber CAFE standards entirely tbrough mix-sbifting. 	The
model 	is 	not 	directly 	applicable 	to 	estimating 	the 	long 	term 	effects 	of
raising a CAFE standard today on automobiles to be produced leveral years
(rom 	DOW. althoulh 	the qualitative 	effects 	discussed 	in Section 	IV 	would
still 	apply. 	It 	is 	presented 	bere 	to 	live 	lome 	examples 	of 	how 	these
effects 	may 	influence an 	estimate of lasoline savings.27 	We 	note 	that 	it
may 	be expected 	that CAFE Itandards effectively let for 	leveral 	years in
the 	future 	may 	bave 	a 	larler 	impact 	on 	lasoline 	livings 	than 	the 	same
standards 	set 	for 	years 	close 	at 	band 	if 	technology 	forcing 	is 	a 	part 	of
lirms' compliance strategies (or (uture years.
Our model 	first 	estimates the changes in 	the composition of the 	fleet
of MY 1989 cars produced under various CAFE standards. 	It then estimates
the changes in gasoline consumed by this fleet and the stock of used 	cars
over 	the 	next 	15 	years. 	Future 	lasoline 	livings 	are 	discounted 	at 	an
annual rate of four percent. 	The model abstracts from possible exogenous
shocks 	in 	such 	areas 	as 	lasoline 	prices 	and 	regulatory 	policy 	that 	may
occur to the neets of cars sold after MY 1989.
Figure 	I 	traces 	the 	lasoline 	consumption 	savings 	from 	progressively
bigher CAFE standards, assuming these standards had 	been put into effect
(or 	MY 	1989. 	The 	dotted 	-No 	Effects· 	line 	traces 	the 	estimated 	savings
under the assumption 	that the sole effect of a 	CAFE standard 	is to raise
the average MPG of MY 1989 cars produced by firms for which the standard
IT The 	figures 	presented 	here 	arc 	not 	meant 	to 	be 	precise estimates
of the effects of any particular CAFE policy that NHTSA may pursue.
13

iJ 	binding. 	It 	therefore 	rails 	to 	account 	for 	any 	of 	the 	four 	effects
discussed 	above. 	The dotted -No Effects· line is always positively .Ioped
and very close to linear (except for the kink between 27.1 	and 27.2 MPG,
wbere 	the 	CAFE 	constraint 	would 	become 	binding 	on 	General 	Motors).
According 	to 	this 	initial 	analysis, 	stricter 	CAFE 	standards 	.enerate 	ever
increasing lasoline uvinlS.
The 	thin 	lolid 	line, 	labeled 	"Market 	Effects,· 	traces 	the 	estimated
uvings 	after 	taking 	account 	of 	the 	-market 	substitution- 	and 	-market
elasticity· 	effecu. 	This 	line 	is 	below 	the 	dotted 	-No 	Effects- 	line 	for
CAFE standards up to 28.4 MPG. 	Tbe reduction in .asoline uvings in this
range 	is 	largely 	driven 	by 	the 	-market 	elasticity- 	effect. 	That 	is, 	this
analysis suggcsts that a binding CAFE standard below 28.4 MPG would bave
acted 	to 	put 	more 	vebicles 	on 	the 	road, 	leading 	to 	more 	gasoline
consumption. 	Above 21.4 MPG, this analysis suggests tbat CAFE standards
would 	have acted 	to decrease the number 	of cars on 	tbe 	road, increasing
tbe amount of gasoline conserved relative to the -No Effects- case.
The 	tbick 	solid 	lint, 	labeled 	-All 	Effects,- 	traces 	the 	estimated
lasoline 	savings 	after 	taking 	account 	of 	tbe 	-scrappage 	effect- 	and 	the
-mileage 	effect: 	as 	well 	as 	tbe 	two 	"Market 	Effects.- 	The 	·All 	Effects"
line is roughly equivalent to tbe "Market Effects- line for CAFE standards
of 	less 	than 	27.6 	MPG 	(CAFE 	increases 	of 	up 	to 	1.0 	MPG 	over 	the
nonbinding 	level). 	In 	this 	ranle, 	tbe 	lasoline 	uvinls 	due 	to 	increased
aman 	car 	scrappage 	rates, 	and 	replacement 	by 	more 	fuel-efficient 	new
UDall 	cars, 	roulhly 	'equals 	tbe 	increased 	consumption 	due 	to 	decreased
larle car scrappage rates. 	Above 27.6 MPG, tbe -All Effects· line is well
below 	the 	~rket Effects· 	line, 	larlely 	because 	in 	tbis 	ranle 	.ufficient
14

Figure 1
- Gasoline Savings From 	CAFE 	Standards
Various Assumptions - 	MY 1989
3
F 	2.6
u
e 	2
I
1.5
S
a 1v
I 0.5n
g 0s
-0.5
26.6
....._...."'... "' ....",.'"....",. "'..... ",. 	'" 	. '" .. ",.
27.1 	27.6 	28.1 	28.6 	29.1
CAFE Standard (MPG)
- 	All Effects 	-f- Market Effects 	,.~ 	No Effects
Fuel Savings from FTC Staff Model In
Billions of Gallons (Present Value)

Dumbers 	of 	.sed 	larle 	cars 	remain 	in 	service 	to 	decrease 	estimated
.asoline savinlS.
Takinl 	account 	or 	aU 	(our 	erfects. 	the 	model 	estimates 	that 	CAFE
standards up to approximately 21.1 MPG (increases or up to 1.5 MPG above
the 	Don binding 	level 	of 	26.6 	MPG) 	(or 	MY 	1989 	would 	cause 	increased
.asoline consumption relative to what would have occurred with a standard
at or below 26.6 MPG. 	The model also estimates that CAFE standards above
28.1 	MPG 	would 	have 	decreased 	aasoline 	consumption 	relative 	to 	a
Donbinding standard. 1I 	Indeed. the model estimates that a 29.1 MPG CAFE
standard 	would 	have 	led 	to 	slightly 	greater 	gasoline 	consumption 	savings
than 	indicated 	by the -No Effects· model, because a good deal fewer cars
would ha ve been on the road.
These 	results 	indicate 	that 	lome 	CAFE 	standards 	may 	not 	reduce
lasoline 	consumption. 	We 	note 	that 	alternative 	policies 	directed 	towards
reducing 	the 	consumption 	of 	gasoline 	(and 	thereby 	reducing 	pollution),
II While 	the 	model 	set 	(orth 	in 	the 	appendix 	estimates 	that 	a
standard 	of 	21.S 	MPG 	would 	have 	reduced 	gasoline 	consumption 	by 	50S
million 	lallons, 	it 	would 	have 	cost 	the 	U.s. 	economy 	an 	estimated 	$9.7
billion 	in 	economic deadweight 	loss. 	SS.3 	billion 	of 	this 	loss 	accrues 	to
U.S. 	consumers 	(with 	the 	costs 	of 	higher 	prices 	on 	large 	cars 	greatly
outweighing the 	benefits of lower prices on small cars) and U.s. producers
lose 	S4.4 	billion 	(as 	losses 	on 	small 	cars 	Ireatly 	outweighing 	higher
profits 	on 	laTle 	can). 	This ,cncrates a 	cost 	to 	society 	of 	approximately
$19.22 	per 	,.Uon 	D"'cd. 	In 	other 	words. 	the 	benefits 	of 	such 	a 	CAFE
policy 	would 	have 	cxceeded 	the 	costs 	if 	the 	negative 	'cxternality
associated with consuming a lallon of gasoline was areater than SJ9.22. 	A
21.5 MPG standard would have also led to a decrease in domestic automobile
iDdustry 	employment 	or 	approximately 	100.000 	jobs. 	Similar 	results
pertain to other CAFE standard levels in this reiion. 	Alain. we Dote that
these 	figures. 	whose 	derivation 	is 	explained 	in 	the 	appendi~ are 	Dot
meant 	to 	be 	precise 	estimates 	or 	the 	crfects 	of 	any 	particular 	CAFE
standard. 	Rather. 	they 	are 	meant 	to 	be 	qualitative 	estimates 	or 	the
impacts or bindinl CAFE standards.
IS

auch 	as 	a 	,alOliDe 	tax 	which 	would 	directly 	discouraac 	coDsumers 	from
uinllasoline. may ,enerate hiahcr lasolinc savinas.~
We 	reiterate 	that 	these 	estimates 	use 	an 	explicitly 	,hort-run 	model
that 	tracks 	the 	,asoline 	consumption 	changes 	caused 	by 	a 	lUsher 	CAFE
atandard 	for 	a 	particular 	model 	year 	close 	at 	hand. 	Standards 	let 	for
leveral 	years 	in 	the 	future. 	if 	they 	can 	be 	credibly 	enforced, 	may 	have
,reater impacts on ,asoline savings. 	We believe, however, that 	this model
,ives 	us 	reason 	to 	conclude 	that 	the 	four 	factors 	discussed 	here 	have
important consequences for 	any estimate of alSoline savings due to CAFE
standards.
VI. Coad•• loa
When NHTSA calculates an estimate of the gasoline savings due to CAFE
standards 	that 	may 	be 	required 	for 	an 	EIS 	or 	for 	other 	regulatory
proceedings, 	we 	urge 	that 	it 	talce 	account 	of 	possible 	effects 	of 	CAFE
.tandards 	on 	(1) 	the 	scrappage 	rates 	of 	used 	vehicles; 	(2) 	the 	tendency
for 	drivers 	to 	alter 	miles 	driven 	for 	different 	fuel-efficiency 	levels;
(3) 	total 	industry 	output; 	and 	(.4) 	producers 	unconstrained 	by 	CAFE
Itandards. 	These 	factors, 	individually 	and 	together, 	may 	serve 	to
significantly 	affect 	the 	potential 	for 	sasoline 	uvings 	for 	CAFE
.tandards. 	Therefore, we believe NHTSA should 	consider these factors 	in
estimating lasoJine savinls from CAFE standards.
• 	See the discussion on this issue in FTC Staff Comments (1918).
16

Technical Appendix·
Andrew N. Kleit
• This 	appendix 	has 	been 	prepared 	by 	a 	staff 	member 	of 	the- Bureau of
Economics 	of 	the 	Federal 	Trade 	Commission. 	It 	does 	not 	necessarily
reflect the views of the Commission or any individual Commissioner.

·-.
I. Deler.,tlo. of A.to.o~lIe Market Model
The automobile market model presented here is the same a. that used
in 	FTC StaCf Comments (1938). 	It 	is an 	extension 	oC the model 	used 	by
Kleit 	(1987, 	1938), 	and 	is 	very 	similar 	to 	the 	simulation 	used 	in 	FTC
Starf Comments (1986). 	Model year (MY) 1987 serves as thc base period.
In 	the 	model 	there 	arc 	five 	types 	of 	automobiles: 	(1) 	Asian 	Basic
Small, 	which 	includes 	reBular 	minicompacts 	and 	subcompacts 	such 	as 	the
Scntra, 	the 	Corolla. 	and 	the 	Hyundai; 	(2) 	Asian 	Luxury 	Sma)). 	which
includes specialty subcompacts and 	reBular compacts such 	as the 	RX7 and
thc 	Stanza; 	(3) 	American 	Basic 	Small, 	which 	includcs 	minicompacts 	and
subcompacts 	such 	as 	the 	Cavalicr 	and 	thc 	Escort; 	(4) 	American 	Luxury
Small. which 	includcs specialty subcompacts and 	reBular compacts. such as
the 	Reliant 	K 	and 	the 	MustanB; and 	(S) 	American 	LarBe. 	which 	includcs
intermediatc 	and 	larBc 	cars 	such 	as 	the 	Cutlass, 	the 	LTD, 	and 	thc 	New
Yorker. 	This breakdown 	is based on cateBories used 	by Ward's 	Automotive
Yearbook. 1 	Luxury European cars. which constitute about 4 pcrcent of the
market, 	are 	excluded 	from 	thc 	model. 	VolkswaBcn 	and 	Yugo 	cars 	are
included in the American Basic small seBment. 	On-shorc Asian production
(autos built by Asian firms 	in the United States) is included in thc Asian
segments. 	·Captivc· 	imports 	(autos 	built 	by 	Asian 	firms 	but 	sold 	under
Amcrican nameplates) are includcd in thc Asian segmcnts.
DurinB the 	1980s. Japancsc car salcs in 	thc 	Unitcd 	States havc 	bcen
restricted 	by 	import 	quotas 	(·voluntary 	restraint 	agrccments·).' 	Currently
thc quota 	is set at 2.3 million 	units. 	Howevcr. during MY 	1987 Japanese
imports 	wcre only about 2.25 	million 	units. 	In 	modcl 	ycar 	1988 Japanese
imports fell to 2.05 million units. as on-shore Japancsc production increased.s
Thus. the initial implicit tariff is set in the model to O.
Each Kament is divided into constrained and unconstrained production.
Constrained 	are 	Japanese 	imports 	(potentially 	by 	the 	import 	quota) 	and
General 	Motors and 	Ford (by CAFE standards). 	Unconstrained production
includes on-shore Asian and off-shore non-Japanese Asian (Korean) output,
1 Ward's Aillomotitlt Yearbook (1988). p. ISS and 233.
2 On-shore Japanesc production is not covered by the import re.traints.
I 	Data received from Oak RidBe National Laboratory.
I

Chryslcr. 	Vol1clwaacn. 	and 	Yugo. 4 	The 	quantities. 	prices. 	and 	fuel.
ctriciencies for each type of car for MY 1987. arc Ihown in Table A-I.
Equilibrium automobile prices and quantities arc computed through a
eeries of rive demand and thirteen lupply equations. 	Quantity demanded is
determined by a act of linear demand curves'
(I) QD. AP + B
where Q is the vector of fivc quantitics. P is the price vector. A is a five
by five matrix of Ilope coefficients. and B is a vector of intercepts.
Quantity lupplied is determined by a set of linear supply curves
(2) QS • C(P·T) + D(P·V) + EP + F
where C is a diagonal five by five matrix of supply coefficients for GM and
constrained Japanese firms. D is a diagonal matrix of supply coefficients for
Ford (with the first two diagonal elements equalling zero). E is a diagonal
five by five matrix of lupply coefficients for the unconstrained firms. F is a
vector of supply curve intercepts. and T is a vector of implicit taxes. T' •
(T l' Tit T I' T4' T I)' 	T is applied to General Motors and offshore Japanese
production. 	T1 and T J are the implicit tariffs for each type of off-shore
Japanese 	car. 	T 1-TJ' 	T ,. T4' T, arc 	the 	implicit 	CAFE 	taxes applied 	to
each type of American car produced by GMe 	V is a vector of implicit taxes
applied to Ford. 	V'. (V l' VJ. Va. V4' V,). where V 1 - 	VJ • 	O. 	The level
of thcse implicit taxes will be generated by the model. 	The model assumes
that OM and 	Ford will choose to meet the relevant CAFE standard rather
than pay CAFE fines.
4 The 	closest 	of 	these 	producers 	to 	being 	constrained 	by 	CAFE
standards is Chrysler. which generally obtains CAFE ratings in the range of
27.5 	to 	28.5 	MPG. 	P.E. 	Godek 	("'The 	Corporate 	Average 	Fuel 	Economy
Standard: 	1978-1988: 	Mimeo, 	Economists 	Inc.. 	Washington 	D.C .• 	January,
1989) indicates that for model year 1989 Chrysler would have approximately
5.6 MPG worth of CAFE credits available. 	For example, if the standard for
model year 1989 had been sct at 27.5 MPG, Chrysler could have had a CAFE
rating as low as 21.9 MPG before it would have run out of credits. 	Chrysler
could 	also 	have 	borrowed 	credits from 	the 	future. 	Thus, the 	model 	will
assume that Chrysler would be unconstrained by any of the CAFE standards
used in Section VI of our comments above.
'With 	the 	imposition 	of 	a 	CAFE 	standard. 	linear 	demand 	curves
lenerate Icss deadweight loss than constant elasticity curves.
• Assume that under one Icenario the implicit tariff on Japanese cars is
SSoo and the implicit CAFE tax is 5300 per MPG for General Moton. 	Using
the formula for calculating implicit CAFE taxes (sec Section III below) and
the MPG per class in Table A-I yields an implicit tax vector T' • 	(500. 500,
300·27.5·«27.5/32.45)-1). 	300·27.5·«27.5/27.42)-1). 	300·27.5·«27.5/25.31 )-1» •
(500. SOOt -1258. 24. 714).
2

CAFE 	standards 	are 	assumed 	to 	be 	just 	nonbinding 	in 	the 	initial
conditions (that iI. that the levels reached by automobile companies under a
lower CAFE .tandard are the .. me as those they would have reached without
any 	CAFE .tandard 	at 	all). 	but 	differentially 	binding 	on 	the 	~il Two"
(General Motors and Ford) if a bindinl standard is enforced. 	Thil is likely
to yield an underestimate of deadweilht loss (OWL), as OWL iI a lunction of
the implicit tax lCluared. 	Crandall and Graham' .uIgest that even a CAFE
standard of 26.6 MPG was bindinl on GM and Ford in MY 1989, meaning
that an implicit tax already applies to these two firms. 	If CAFE standards
are imposed, they are assumed to be binding, and the implicit tax per ·Big
Two· car is calculated accordinlly. 	The system of 21 equations (five demand
curves, 	thirteen 	supply 	curves, 	two 	CAFE 	constraints, 	and 	one 	import
constraint) 	in 	21 	unknowns 	(five 	prices, 	thirteen 	quantities, 	and 	three
implicit taxes) is solved and the implicit tariff and the shadow tax per MPG
lor GM and Ford are iterated until the desired quota and CAFE standard
level are reached.
Based on past studies of the demand for automobiles we assume that a
10 percent increase in the small car price will lenerate a 20 percent decline
in the quantity of small cars demanded and that a 10 percent increase in the
price of large cars will 	lower the quantity of large cars demanded 	by 	30
percent.'
The point elasticities of demand at the original 	1987 equilibrium are
shown in Table A-I. 	The own-price elasticity of demand for automobiles is
assumed 	to 	be 	one. (This 	is consistent 	with 	the 	results 	reported 	in 	Irvine
(l983).~ 	The 	cross-elasticities shown 	should 	not 	be 	interpreted 	as 	precise
ligures, but merely internally consistent with the overall market demand and
the own-elasticities for each of the segments which range from 2 to 3. 	The
method for the derivation of the cross-elasticities is available upon request.
To 	our 	knowledge, 	no 	study 	exists 	of 	short-run 	cost 	curves 	for
automobile 	production. 	Results obtained 	by 	Friedlander el QI.IO indicate
'f Robert W. Crandall and John O. Graham, "The Effect of Fuel Economy
Standards on Automobile Safety: JournQI of lAw Qnd Economics 32:2 (April,
1989), 97-118.
• These sensitivity 	(or 	elasticity) estimates 	are 	those 	found 	in 	James
Langenfeld 	and 	Michael 	Munger, 	"The 	Impact 	of 	Federal 	Automobile
Regulations on Auto Demand,· unpublished draft, Federal Trade Commission
(June 	1985) and 	were those 	used 	in FTC Staff Comments (1986) and FTC
Staff Comments (1988).
• F. 	Owen 	Irvine 	Jr .. 	·Demand 	Equations 	for 	Individual 	New 	Car
Nodels.· Souturn Eco1fomic Jour1fQ1 (1983), 764·782.
10 A. F. Friedlander, C. Winston, and K. Wang, ·Costs, Technology, and
Productivity 	in 	the 	U.s. 	Automobile 	Industry,· 	Bell 	Jour1fQ1 01 Economics
(1982). 1-20.
3

that the iDdustry .ay have constant lonl run marlinal cost curves. 	In the
abort-run, however, it seems likely that marlinal costs are increasing. 	Tbus,
the point elasticity of supply (mar,inal cost) in tbe model is set equal to 2
ror 	MY 	1919. 	Tbis 	anumes 	tbat 	wbile 	tbe 	industry 	has 	a 	competitive
structure, tbere are short term economic rents to be earned in the laic of
automobiles.
We assume that witbout hilher CAFE standards General Motors would
.ave reacbed 27.2 MPG in MY 	1989, while Ford 	would have reached 26.6
MPG.ll 	Thus, if the MY 1989 CAFE standard had been let at 27.5 MPG, the
model assumes that meeting that standard would have required a -stretch" of
0.3 MPG for GM in MY 1989 and 0.9 MPG for Ford.
II. DerlntloD of ClloliDe CODluDlptioD CbaD,el
To 	measure 	the 	change 	in 	,asoline 	consumption 	over 	time 	that 	will
result 	from 	changing 	the 	CAFE standard 	in 	MY 	1989, 	it 	is necessary 	to
estimate and compare (J) the lifetime ,asoline consumption of new cars sold
under the standard, (2) the estimated gasoline consumption of the cars that
would bave been produced had tbe bigher CAFE standard not been imposed,
and 	(3) 	tbe -scrappage effect" (the 	change 	in 	the stock 	of 	used 	cars 	that
results from a change in the price of new cars) to determine the total stock
of cars in operation.12
Data 	on 	miles 	driven 	and 	scrappage 	rates 	are 	incorporated 	into 	the
lasoline 	consumption calculations used 	in the estimates represented 	by 	the
solid eyotal 	Effects" -line of Figure 	1.13 	The scrappage rates are adjusted
for 	new 	car 	price 	changes 	using 	Gruenspecht's 	estimates. 14 	Gruenspecht
showed 	that 	if 	the 	price 	of 	new 	cars 	is 	raised 	(lowered), 	it 	causes 	a
significant decrease (increase) in the scrappage rates of used cars. 	Here we
assume tbat Gruenspecht's results can be applied to each of the three classes
11 NHTSA 	Final 	Rule, 	53 	Federal 	Register 	39282. 	While 	the 	model
requires such a -starting point", the results of Section VI are not sensitive
to which particular starting point is selected.
12 Several 	studies, 	such 	as 	Howard 	K. 	Gruenspecht, 	-Differentiated
Regulation: 	A Theory 	with 	Applications to 	Automobile Emission 	Controls,"
Yale 	Univcnity 	Pb.D. 	dissertation, 	1982, 	and 	Gruenspecht, 	"Differential
Regulation: 	The 	Case 	of 	Auto 	Emissions 	Standards," 	American 	Economic
Re,iew 	(1982), 	328·331, 	have 	found 	tbat 	scrappage 	rates of 	used 	cars 	are
ai,nificantly affected by new car prices. 	See also .James A. Berkovic, "New
Car Sales and Old Car Stocks: RAND Journal of Economics (J 985), 195·214
and Richard W. Parks, -Determinants or Scrappage Rates for Postwar Vintage
Automobiles,- EcoPlOmetrica (1977) 1099·11 IS. 	-
U 	Fi,ures obtained from the Motor Vehicle Manufacturer 'Auociation,
Motor Yehicle Ft:lCts IJM Figures (1987).
14 Gruenspecbt (l982b).
4

of automobiles (Basic Small, Luxury Small, and Larle). 	For purposes of the
...oline 	consumption 	modcl, 	Japancse 	cars 	are 	combincd 	....ith 	their
correspondina Amcrican .amcnts. 1S
Many 	of 	thc 	chanles 	in 	fleet 	composition 	rcflcct 	larae 	car 	buycrs
awitchinl to 	new small cars in 	response 	to changes in 	rclative car prices.
But because smaller can use less fucl, the marginal cost of drivina declines,
and drivinl is encouraled. 	The model uses Blair ~t G/.'S18 findinls to adjust
for 	chanles 	in 	the 	rate 	of 	use 	of 	new 	cars 	induccd 	by 	hilher 	CAFE
standards for the estimate described by the thick solid line of Figure 1.
For the most part, the chanles in aasoline consumption idcntified here
do 	not 	occur 	in 	the 	model 	years 	to 	....hich 	the 	CAFE standard 	would 	be
applied. 	Rathcr, they are the summation of aasoline consumption changes
for the ensuing IS years after a one ycar rise in the binding CAFE standard.
A real discount rate of 4 percent is used.
III. Tbe Mathe.atici of Har.oalc Ayera.IDI
By statute, CAFE standards rclate to the ·harmonic· averale of a firm's
vehicle 	fuel-efficienc)'. 	This section 	will 	discuss how 	harmonic averaging
affccu the implicit t.xes and subsidies lenerated by bindinl CAFE standards
as well .s the relative impact of harmonic versus ·simple· averaging.
Assume that • 	firm 	makes only 	two types of cars; a 	larle relatively
fuel-inefficient 	model, 	.nd 	a 	small 	relatively 	fuel-efficient 	model. 	As
discussed 	in Section 	III of the comments above, the explicit fine. 	F, on 	a
firm for failinl to reach the CAFE standard is cqual to
if the firm does not reach the standard, ....here S is the level of the CAFE
.t.ndard, QL .nd Q. arc thc number of I.rge .nd small c.rs sold by the
firm. and MPG is thc firm's .verAlC rl:lct-efficiency.
11 The 	MPG 	valucs 	for 	cach 	of 	the 	thrce classes can 	be 	determined
from the information used in the automobile market model. 	The entire fleet
fuel intensity for MY 1973 ....as about 14.2 MPG. 	The fleet fuel intensity for
MY 	1987, ....hich Cln 	be calculated from Table A-I, .....s 28,49 MPG. 	The
model .ssumes that the 	ratio of fuel 	intensity between classes is the same
for cach year. 	With this assumption, knowledle of the fraction of c.rs in
- 	each 	clan for 	1973, aDd tbe cntire (Jeet (Dcl intensity (or 	1973, the 	fuel
iatcnsity 	for 	each 	cl.ss 	of 	Dew 	car 	in 	1973 	caD 	be 	cstimated. 	We 	also
aavme 	th.t, for 	each 	cl.ss of Clr, the 	relcvant MPG 	Irew at a constant
...te bet....een 	1973 .nd 	1987. 	MPG's are then calculated .ccordill,ly. 	The
I.el intensity of cars produced before 1973 is assumed equal to be to the
1973 level.
II Blair, Kaserman, .nd Tcpel (1914).
s

The measurement of a firm's CAFE level was not defined as the simple
averale of a manufacturer's fleet MPG. 	Instead, a firm's CAFE level is the
laarmonic averale of that firm's fleet MPG.l1 	The harmonic averale for the
firm is calculated by
where ML and Ma are the fuel-efriciencies of the two types o~ ~n.
Using the 	harmonic average. the marginal CAFE fine 	to the firm 	of
producing a car of type J is
(5) 	dF/dQl. 50·(S-2MPG+(MPG'/Ms»
Assume 	DOW 	that 	the 	standards 	are 	binding. 	In 	that 	case 	MFG-S. 	the
explicit fine of SSO per MFG per car is replaced by a shadow tax L and the
implicit CAFE tax on a car of type J becomes
where L is the value of the constraint discussed above.
The marlinal fine derived above presents a more difficult problem to
manufacturers than would occur with a standard based on simple averaging.
Consider a firm that is deciding whether or not to produce an additional car
with fuel-efficiency equal to 20.0 MPG where the binding CAFE standard is
27.S MFG. 	If simple averaging were used. the firm 	would 	have to offset
that additional unit by producing one car with fuel-efficiency of 3S.0 MFG
(or the equivalent). 	Under harmonic averaging, however, to produce another
unit of 20.0 MFG. the 	firm 	must also produce the equivalent of one unit
with fuel-efficiency of 44.0 MFG. 	Thus. compared to simple averaging. the
harmonic avera.ing used makes the CAFE standard more difficult to meet.
If Public Law 46:15-2003. 	One property of a harmonic avenle is that
if it is doubled. fuel consumed by driving the same number of miles in each
type of car is halved.
6

Table A-I
Para.eten Used I. CAFE SI.ulatloD
Demand Elasticity Table
Class 	I
I) Alian Small 	-2.000
2) Alian Luxury Small 0.217
3) Domestic Small 	0.856
-t) Domestic Lux. Sma. 	0.1 65
5) urle 	0.0 15
2
0.243
-2.500
0.446
0.539
0.083
3
0.334
0.125
-2.000
0.103
0.010
4
0.355
0.837
0.583
-2.500
0.103
5
0.704
2.661
1.160
2.237
-3.000
Class
1
2
3
4
S
1987 Totals by Class
Price 	Quantity 	MPG 	Cars/Job
(Initial) 	(lnit.)
(SOOO) 	(million)
8.689 	1.748 	35.51 	22.65
13.764 	1.173 	29.57 	19.38
8.373 	1.168 	32.45 	7.55
10.719 	1.884 	27.42 	6.46
15.077 	3.645 	25.31 	5.40
Initial Quantities by Firms
(millions of units. 1987 production)
Constrained Uncstr.
Class 	GM 	Ford 	Chrysler 	Other 	Asian 	Asian
I 	0.000 	0.000 	0.000 	0.000 	1.343 	0.405
2 	0.000 	0.000 	0.000 	0.000 	0.909 	0.264
3 	0.4J6 	0.461 	0.150 	0.140 	0.000 	0.000
.. 	0.184 	0.466 	0.534 	0.000 	0.000 	0.000
5 	2.234 	1.128 	0.283 	0.000 	0.000 	0.000
Supply Elasticity: 2.0 (all firms and classes)
Source 	for 	prices: Ward's 	.Alltomoti.,~ Y~arbook. 	1988. 	pp. 	216-221 	and 	287-
293. 	Source for quantities and fuel-efficiencies: P. S. Hu and L. S. Williams,
-Lilht Duty Vehicle MPG and Market Shares Report: 1st Six Months Model
Year 1988: Oak Ridge National Laboratory. 1989. E-41 to E-44.
7