Demandfor Electric Vehicles

in Hybrid Households:
An Exploratory Analysis

Kenneth S. Kurani
TomTurrentine
Daniel Speding

Working Paper
UCTCNo. 232

The University of California

TransportationCenter
Universi~¢of California
Berkeley, CA94720
The University of California
Transportation Center

The University of California Center activities. Researchers

Transportation Center (UCTC)
is one of ten regional units
mandated by Congress and
established in Falt 1988 to
support research, education,
and training in surface trans-
portation. The UC Center
serves federal Region IX and
is supported by matching
grants from the U.S. Depart-
ment of Transportation, the
California Department of
Transportation (Caltrans), and
the University.
Based on the Berkeley
Campus, UCTCdraws upon
existing capabilities and
resources of the Institutes of
Transportation Studies at
Berkeley, Davis, Irvine, and
Los Angeles; the Institute of
Urban and RegionaI Develop-
ment at Berkeley; and several
academic departments at the
Berkeley, Davis, Irvine, and
Los Angeles campuses. o magazine presenting sum-
Faculty and students on other
University of Caiifomia
campuses may participate in
at other universities within the
region also have opportunities
to collaborate with UCfaculty
on selected studies.
UCTC’seducatior.’d and
research programs are focused
on strategic planning for
improving metropolitan
accessibility, with emphasis
on the special conditions in
Region IX. Particular attention
is directed to strategies for
using transportation as an
instrument of economic
development, while also ac-
commodatingto the region’s
persistent expansion and
while maintaining and enhanc-
ing the quality of life there.
The Center distributes reports
on its research in working
papers, monographs, and in
reprints of published articles.
It also publishes Access~ a
maries of selected studies. For
a list of publications in print,
write to the address below.

University of Califorma
Transportation Center

108 NavalArchitecture Building

Berkeley,California 94720
Tel: 510/643-7378
FAX:5101643-5456

Thecontentsof this report reflect the viewsof the authorwhois responsible

for the facts andaccuracyof the dampresentedherein. Thecontentsdo not
necessarilyreflect the official viewsor policiesof the State of Californiaor the
U.S.Department of Transportation.This report does not constitute a standard,
specification,or regulation.
Demand for Electric Vehicles in Hybrid Households:
An Exploratory Analysis

Kenneth S. Kurani
TomTurrentine
Daniel Sperling
Institute of TransportationStudies
Universityof California at Davis
Davis, CA95616

WorkingPaper
May 1994

UCTCNo. 232

TheUniversityof CaliforniaTransportationCenter
Universityof Californiaat Berkeley
ABSTRACT
Previousstudies of the potential marketfor battery electric vehicles (BEVs)have
reached contradictory conclusions. Whatthey share are untested or implausible assumptions
about consumerresponse to newtransportation technology. Weframe the BEVpurchase
decision in terms of a household’sentire stock of vehicles, car purchasebehaviorand travel
behavior. Within this framework,householdswhich ownboth electric vehicles and gasoline
w~hiclesare called "hybrid households".Becausenearly all consumersare unfamiliar with the
characteristics of BEVs,wedesigned an interactive interview based on week-longtravel
diaries, which we call Purchase Intentions and RangeEstimation Games(PIREG)to explore
hypotheticaJ hybrid householdvehicle use. Our primary finding is that consumers’perceived
ctriving range needs are substantially lower than previous hypothetical stated preference studies
conclude. Wef’md evidence of a viable market for BEVswith 60 to 100 miles driving range.

Demandfor electric vehicles

Keywords:electric vehicles; travel behavior; newtechnologies

INTRODUCTION

Electric vehicles promiselarge energyand environmentalbenefits (Sperling; 1994)o

Theyare quieter and will likely have lower operating and maintenancecosts than internal
combustionengine vetficles (ICEVs).Electric vehicles will also be capable of being recharged
at home,workand other unconventional, but convenient, locations.
At least initially, electric vehiclesare usingbatteries to store electrical energy.But
batteries have low energydensity, whichresults in greatly reduced driving ranges. Also,
typical battery recharging times are measuredin hours, not minutes. Limited range and long
recharge times create uncertainty and skepticismaboutthe possibifity of selling battery electric
vehicles (BEVs)to consumershabituated to long driving ranges and quick, ubiquitous
refueling. Conservativesales estimates in turn lead to high cost estimates becausecosts are
spread over few vehicles. Highcost estimates iteratively reinforce estimates of a minimalBEV
market.
In a broad sense, the initial target marketsfor BEVsare commercial,utility and
governmentfleets and the growingnumberof multi-car households. Wefocus on the
household market in this paper. Accordingto the 1990 NationwidePersonal Transportation
Survey, 58 percent of householdsin the United States (a total of over 54 million) have two
morevehicles (Hu, et al, 1992). BEVsshould not be seen as simple one-for-one substitutes for
ICEVs.BEVsoffer significant limitations as well as newcapabilities. BEVscomprisean
alternative travel technologywhichhouseholdsmustlearn to integrate with familiar gasoline
vehicles° AhouseholdwhichcombinesBEVsand gasoline vehicles in its stock of vehicles is
one exampleof what wecall a hybrid household. In contrast to a hybrid vehicle which
combineselectric and heat engine drive systems in one vehicle, a hybrid householdchooses
two vehicles with different types of energy systemsand then must allocate householdtravel
accordingly.
In manyways, the introduction of BEVsto the car marketis analogousto the
inlxoduction of microwaveovens to the oven market: a classic exampleof an appliance with an
entirely newset of functional features, whatBerkowitz,et al (1992, p.290) call a "high
learrfing product". Sales of microwaveovens were slow but steady. Marketpenetration
increased from 15%of US. households in 1980 to 75-80%by 1992 (ibid). Initially,
consumerswere concerned whether microwaveovens cooked food as tastefully as did
conventional ovens. To convince consumersof the potential uses of microwaves,special
"cooking classes" were given to demonstrate the technology. Consumersdiscovered the
comparisonto heat ovens was misplaced; the microwaveoven did not displace the traditional
heat oven. Instead, it becamean additional oven, whichresulted both in reallocation of
previous cookingtasks and newcooking practices. Householdswhichinitially rejected
microwaveovens because of perceived inferior cooking characteristics, nowuse microwave
ovenseverydayfor entirely new"cooking"tasks such as defrosting foods, re-heating coffee or
cooking any of hundreds of products which have been specifically repackaged for microwave
ovens.
As happenedwith initial microwaveoven market research, electric vehicle market
research often fails to identify a marketfor this "radical" newtechnology.Wewill showthat
this inferred lack of a marketfor BEVsis an artifact of previous research designs. Consumers
questioned in previous studies had neither enoughinformation nor experience with the
technology to offer informedopinions or to have formedpreferences. Consumers’lack of
experience with limited ranges, homerecharging and other features of BEVsrenders
conventional survey methodsof limited usefulness, both in predicting sales of BEVsand in
providing explanations and insights as to whichhouseholds will consider BEVpurchases.
Oneresearch design whichappears particularly motivatedto find householdslikely to
consider BEVs,is in fact too narrowlyfocused. Thesestudies focus on identifying "green"
consumerswilling to pay considerably morefor a zero emission vehicle. But these studies are
often flawed because they allow environmentallymotivated consumersto initially respond
positively to BEVswithoutcarefully consideringthe lifestyle implications of limited range.
 Conversely, they exclude from the BEVmarket less environmentally motivated households.
But, as wewill show,after somereflection on their lifestyle wantsand needs, these
householdswhichare excluded from the "green" market, mayfind that short ranges are not a
real limitation and that other features of BEVssuch as homeor workplacerecharging make
BEVsa superior lifestyle choice comparedto gasoline vehicles.
In a mature market, in whichhouseholdscan draw on the experience of other
consumersto help evaluate unfamiliar products, these lifestyle choices are readily made.But in
the case of electric vehicles, a maturemarketis years away. Absentthe informationand product
familiarity whichcharacterize maturemarkets, consumerresearch must do its best to enhance
the informationcontext of survey work,creating learning situations as well as exploring with
householdsthe lifestyle implications of newtechnology. To provide this decision context, we
developed PIREG(Purchase Intentions and RangeEvaluation Games),an interactive lifestyle-
preference interview. PIREGwas designed to provide the necessary information context for
householdsto evaluate the strategic impacts of driving range limits and recharging
characteristics of BEVs.
Weconducted 51 PIREGinterviews in three metropolitan areas of California. The
households chosen for this study represent potential hybrid households. Theyowntwo or
morecars, and thus potentially havesomeflexibility in assigning different vehicles to different
drivers and trips. Theybuy newcars. Mostownsingle-family homeswith a garage in which
they park at least one vehicle. Weincluded a fewrenters to explore issues aroundresidential
tenure and availability of dedicated parking for recharging a BEV.
Theresults of this research provide an in-depth viewof the decision variables and
strategies householdsconsider whenevaluating BEVs.Basedon these results, weidentify
three types of households: those whoare pre-adapted to BEVs,those whocan easily adapt,
and those for whom adaptations are difficult or impossiblegiven their routines, lifestyle goals
and budgets. Weexplore the vehicle driving ranges that householdsare willing to consider and
potential impacts of BEVson their travel behavior. In short, we explore what it might meanto
becomea hybrid household.

THREE TYPES OF PREVIOUS BEV MARKET RESEARCH:

INADEQUATE APPROACHES AND A CONTRADICTION
Previous research on BEVshas focused on predicting the size of the market at the
expense of understanding market dynamics for a fundamentally newconsumerproduct. Many
of these studies have relied uponconvenientrather than appropriate data samples. Others have
used data fromhypothetical choice exercises. Almostall, we believe, rely on an implausible set
of assumptions regarding consumerbehavior. Such shortcomingsexist precisely because there
are no sales data for BEVs.In the absenceof sales data, researchers have tried three methodsto
developestimates of BEVmarketpotential -- attitude studies, travel behavior analyses, and
stated preference surveys.
Thesethree research streams present an apparent paradox. Attitude studies and travel
behavior analyses tend to showBEVsto be a practical and desired technology, but stated
preference studies typically conclude consumersare unwilling to consider BEVsat anything
but "fire sale" prices. This paradoxcalls for close scrutiny of the methodsand findings in
these studies.

Attitude Surveys
A numberof attitude surveys and somefocus group studies by auto manufacturers,
electric utilities and auto marketanalysts havefound, rather universally, a sizable percentageof
consumerswhoare interested in, and favor, electric vehicles and other alternatives to gasoline
(Buist, 1993; PG&E, 1993; Fairbanks, Maulinand Associates, 1993;). It appears that electric
vehicles in particular have a special fascination over other propulsion systemsbecausethey
have the most progressive imagetechnically and environmentally(Turrentine, et al, 1992).

2
However,these attitudes are far removedfrom vehicle purchase and use; they represent the
ideals of consumersand not their full decision process. Additionally, these studies often report
conflicting attiVades. Theyreport that on the one hand consumersstrongly favor electric cars,
but on the other, wantsimilar driving range as their gasoline vehicles. Suchconflicts often
lead to simple dismissal of consumerattitudes as merely "feel good"answers.
Anotherflaw is that attitude studies whichset out to find a "green" marketare unduly
constraining their search for a BEVmarket. Ford MotorCo. (Buist, 1993) has reported using
this approach;first, finding the environmentalconsumer,and then culling those willing to pay
the purchase price premiumFord projects for BEVs.This approach maybe interesting to
manufacturersfor several reasons. It captures those with certain strong convictions about
BEVs;it mayidentify someconsumerswhoare willing to pay morefor an BEVthan a gasoline
w;hicle; and it mayeven identify consumerswhohave not previously purchaseda newvehicle,
but might buy a BEV.But manyof those with strong environmental convictions have neither
appropriate vehicle use nor purchasebehavior. Byfocusing on the green aspect as an initial
filter of the market, studies such as Ford’s eliminate a wide set of consumersfor whomBEVs
offer practical advantagesas part of a hybrid householdfleet. Wehave foundin previous
st~:~dies (Turrentine,et al, 1992)and in this workthat broaderlifestyle issues are better primary
filters for the BEVmarketthan envh’onrnentalconvictions.

Travel Behavior Studies

In studies based on travel behavioror constraints analyses, researchers have largely
focused on the issue of limited range. Typically such studies attempt to count households
wlhich have two or morevehicles and travel habits which can accommodate a limited range
BEV.The primary assumption in these studies is that potential BEV-owning households must
halve at least two vehicles. Theother common assumptionis that there can be no pattern of
d~y vehicle use in whichboth vehicles travel beyondthe range of the BEVoThe data used in
these studies often comefrom the NationwidePersonal Transportation Survey (NPTS)or the
AmericanHousingSurvey (AHS). The NPTScontains a one clay travel diary. The AHSasks
only about typical travel and commute travel. For examplesof these constraints analyses, see
Deshpande(1982), Kiselewich and Hamilton(1982) and Nesbitt, et al (1992)°
Theprimaryfinding of most of these studies is that morethan 90 percent of two car
householdscould use one vehicle with 1130miles of daily range and that most"second" cars are
used morethan 1{30miles on only a few days per year. In general, such studies concludethat
55 to 60 million householdscould accommodate a 100 mile range vehicle. Studies of this type
in other countries return equally impressivepotential marketestimates. A study conductedin
the former Federal Republic of Germanyconcluded 7 million households could accommodate
BEVs(Hautzinger, et al, 1991).
Oneof the morerecent of these studies adds a further constraint on top of vehicle
ownershipand use -- the householdmust have a logical place to recharge the BEV.Theyfind
about 28%of Americanhouseholds (28 million households) could accommodatea BEV
(Nesbitt, et al, 1992). Greene(1985) uses the travel behaviorapproachbut distinct data;
analyzes multi-dayrefueling diaries, and infers underlyingdistributions of travel. Heconcludes
that with 95%probability, half of all householdvehicles travel less than 105 miles per day on
95%of all days. There wasno substantive difference betweenvehicles in single- and multi-car
households.
A recent study by GeneralMotors, aimedat understandingthe market for electric
vehicles, concurs that the majority of any household’stravel requires minimalrange or
passenger payloads (Dables, 1992). Potential BEVownerskept three-week driving logs in that
study. GMreported 84%of their sampledrove less than 75 miles a day and in only 5%of trips
were morethan two persons in the car.
All these studies present reassuringlylarge marketpotentials. But the limitation of the
travel behavior approachis that it doesn’t measureconsumerpreferences. Whilemeasuringa
"potential market", these studies don’t examineattitudes or social processes whichwill shape
consumerlifestyle choices. Additionally, they analyze vehicle stocks rather than newcar sales.

3
Skeptics of the potential marketfor BEVshave criticized constraints analyses, arguing that
regardless of howpeople actually use their vehicles, consumersprobably won’t give up
unlimited range or fast refueling of ICEVs.Hamiltoncomplainedthat such studies were merely
wishful thinking (Hamilton, 1983); the third approachto BEVmarket studies, stated preference
techniques, appear to support this argumentquite forcefully.

Stated Preferences
Stated preference surveys present consumerswith options, in written or telephone
survey form and ask them whichthey wouldbe willing to buy. Each option is described by
attributes common to all the possible choices. Theattribute levels are varied over severn trials
to elicit different choices of vehicle options. Econometricmodelsbased on such data assign
partial utility valuesto consumer preferencesfor vehicleattributes. Thepartial utilities for
driving range are often then used to estimate a purchaseprice penalty for limited range vehicles.
Virtually every stated preference study estimates hugeaveragepenalties for limited
range vehicles. For exampIeconsider the estimated average purchaseprice penalty (adjusted to
19915)assigned to a 50 mile range vehicle whencomparedto a 200 mile vehicle for the
following three studies: Morton,et al (1978), $10,000; Beggsand CardelI (1981), $16,250;
and morerecently, Bunch,et al (1993), $15,000. In a slightly different study, Calfee (1985)
calculates individual-specific price penalties. Therange of estimatedpenalties is large, but
manyare close to the average ~enaIties reported in the other studies -- even for consumerswho
claimed they preferred BEVs..Considering that the average price of a newautomobile in
1991 was $16,700 (MVMA, 1992), these studies suggest that on average consumers would
indifferent to the choice betweentwo cars whichwere identical, except one wasfree and had a
50 mile range, and the other, for whichthey must pay full price, had a 200 mile range. Using
these large average penalties for limited range, projected BEVsales are very low. Market
penetration estimates in these studies range from 2%downto 0%.
Weare skeptical regarding this conclusionfor tworeasons. First, the averageutility is
irrelevant to the dynamicsof market development.The average penalty for limited range makes
an apparently compelling argumentfor those opposedto the introduction of BEVs.But
"average" consumersare not, by definition, the first buyers of somethingnew. It is the
distribution of disutilities whichmatters. Theappropriate objective is to determinehowmany
consumershavepositive, or relatively small negative, utilities for BEVs,not the magnitudeof
the average utility. Oursecond reason for skepticism is the underlying assumptionsregarding
consumerbehavior in stated preference studies and the contradictions to these assumptionswe
find in our work. Weaddress these issues next.

Volatility in stated preferences for range

The underlying assumptions about consumerbehavior contained in econometric models
based on stated preference experimentsseemuntenable to us. A completecritique is provided
elsewhere(Turrentine and Sperling, 1991). Here we focus briefly on the characteristics
preferences. In order to makeinferences about the value placed on driving range, it must be
assumedthat respondents have welt formedpreferences for range. Preferences have specific
properties, e.g. transitivity and communativity.Mostimportantly for purposesof forecasting
future market shares, preferences must be stable or there must be enoughlongitudinal data and
an adequatetheoretical understandingto also forecast the rate of changeof preferences. These
are highly speculative assumptionsfor attributes with whichconsumershave no experience. As
wewill show, consumer"preferences" for driving range shift dramatically based uponsmall
incrementsof information. Suchshifts are evidenceof instability and mayresult in non-
transitivity of "preferences"for different attributes.
Not only do stated preference studies find very high penalties for limited range, but so
too do other less rigorous, studies such as those of Ford (Buist, 1993) and J.D. Powers
(1993). Focus group studies (Mortonet al 1978; J. D. Powers1991) report that consumers
state their desireddriving rangeto be essentially that of their gasolinevehicle.
 Suchresponses are groundedin consumers’past experience with familiar vehicles.
"[’he contrast betweenthese responsesand the actual daily distance provide the motivationfor
tiffs PIREGinterviews. Webeganto explore this problemin a previous study in which11
focus groupswere conductedwith participants of a drive test clinic of electric, compressed
natural gas and methanolfueled vehicles in 1990in Pasadena,California (Turrentine, et al,
I’992). In the focus groups, weelicited initial estimates of neededdriving range fromeach
participant at the start of the session. Thenwediscussed range needs in a numberof ways,first
asking participants to estimate their actually daily driving, and then to maketrade-offs between
range, fuel prices and vehicle prices to explorethe stability of their initial rangeneedestimates.
Theprimaryfinding wasthat participants’ stated preferences for range wereextremelyvolatile
and changeddramatically under the influences of newinformation, attitudes expressed by other
group members,and attempts of the moderator to influence responses by suggesting range
re, lated problems.Somerespondents’stated needs increased, but overall, there wasa pattern of
drastic reductionsin stated daily rangeneeds.
This finding suggested there wasa learning curve for driving range. Whileour sample
wassmall and the setting informal, we found nothing to support the extremeaverage penalties
reported in stated preference work. Weconcludedthat innovative survey and interview
methodswere needed to provide both consumersand researchers an adequate context to
understand and measure potential consumerdemandfor range.

A NEW APPROACH

Previous studies of potential BEVdemandcometo very different conclusions: travel

behaviorstudies find that householdscould easily use a limited range vehicle, while stated
preference studies find limited range to be an overwhelmingdrawback. Someconsumer
attitude studies showpeople view BEVsvery favorably. Yet these studies have also been used
to showthat those consumerswhoshould view BEVsmost favorably, so called "green"
consumers, are unwilling to buy BEVs.Wesuggest the need for a newapproach, one which
provides consumerswith moreinformation about BEVrange and recharging characteristics and
attows themto reflect on their owndriving and vehicle use patterns. To understandpotential
BEVmarkets, weinvestigate in detail the travel and vehicle purchasedecision behavior of
households,as well as their lifestyle plans. In order for both weand our study participants to
gain adequateinformation, weemployinteractive stated lifestyle-preference techniques.
Interactive stated lifestyle-preference techniquesdiffer fromstated preference
techniques in that the researcher and participants engagein simulateddecision makingcontexts
designed from actual behavior of the household. The process creates hypothetical choice
situations constructed from the real material of a household’sactivities. This process increases
the validity of responsesand allows the researchers to observethe lifestyle decisions of the
household. Previous examplesof this type of workinclude the HouseholdActivity
Transportation Studies (HATS)developedby Peter Jones (1979) for studies of transportation
modechoice, and the Car Use Patterns Interview Games(CUPIG)developed by Martin Lee-
Gosselin (1990) to study householdresponses to motor fuel rationing.

PIREG(Purchase Intentions and Range Estimation Games)

Withthe participation of MartinLee-Gosselin,wedevelopedan interactive stated
lift, style-preference interview, PIREG,to explore householdresponses to limited-range, home-
rechargeableBEVs.In PIREG,like other interactive lifestyle-preference transportation studies,
participants keepone weekdiaries of all householdmotorvehicle travel. Thesediaries are
transferred to a time-line whichcontains completetrip-by-trip informationon trip purposes,
origins, destinations, distances, start and end times. This chart is usedto construct "whatif"
situations in a two hour interview with participants in their home.Thechart and other travel
information about the householdare used to see if they understandthe problemsand questions
posedin the interview.

5
 A PIREGinterview has five parts:

CALIBRATION: The household is questioned about the typicality of the week on

.
the chart and about daily, weekly,monthlyor annual travel not foundon the chart
(including travel by other modes).

MINIMUM RANGE ESTIMATION: One of the household vehicles is "replaced"

.
by an exact copy, except the copy has a limited range and is recharged at home.
Theinterviewers choose a "challenging" range - a range whichwill cause some
problemsin completing the weekof actual travel. The household membersreview
the weekand discuss howto solve the probtemspresented. If they successfully
solve these problems,a lower range is proposed. Theprocess is repeated until the
householdarrives at the lowest range vehicle it is willing to accommodate.
3~ ADAPTATION: Using the mivSmumrange vehicle, the interviewer presents the
householdwith further problemsnot contained in the diary, such as a medical
emergencyor other real situations whichmight demandmorerange or further
behavioral change. The household continues the process of problemsolving.

OPTIMIZATION: The household is asked to optimize the use of their hybrid fleet
.
of vehicles under the condition of higher operating costs for their gasoline vehicle,
using the disincentive of a very high gasoline price.

° COMFORTABLE RANGEESTIMATION: Finally the household is presented

with a priority evaluator (PE) table of driving range and recharge times. Each
option has an associated price. Householdschoose to "buy" a newrange/
rechargingpackageor to retain the one they arrived at in the adaptation phase.

PIREG Sample
Rather than survey the general population, for PIREGwe sampled from households
whoare buying newmotor vebdcles in California. Weselected households whosepast vehicle
purchase and lifestyle patterns makethem probable candidates for BEVs.These households
owntwo or morevehicles, all of whichwere reliable and suitable for long distance travel, and
the last newvehicle they purchasedwasa four cylinder sub-compact,compact,mirfivan or
small pickuptruck -- the bodystyles most likely to first be offered as BEVs.This allows us to
easily reviewthe lifestyle decisions whichled to the purchaseof that particular vehicle. The
households are predominately ownersof their ownsingle-family residence and most have
household incomesgreater than $50,000. Public studies on the annual newcar buying market
do not exist; howeverestimates based upon a Newsweek study of newcar buyers indicate
there maybe 200-350thousand California households per annumin the population defined by
our PIREGsample (Newsweek1990). This estimate is substantiated by Rutherford et
(1994) whofind 30%of the annual car market fit the samecriteria as the population from
which we drew our PIREGsample.
Fromthis popuIationwe recruited a variety of householdtypes to explore distinct
lifestyles, including the following: householdswith various numbersand ages of children,
including teenage drivers; households with one and two wageearners; and households of
retired persons. WeconductedPIREGinterviews in three areas of California -- Sacramento,
Santa Clara (southern San Francisco Bay area) and Orange(southern Los Angeles area)
counties -- to uncoverany important contrasts in infrastructure, urban form and consumer
informationrelated to BEVs,~ limited range and lifestyles. All three of these counties are largely
suburban,reflecting the dominantlifestyle choice of Californians.
RESULTS FROM THE PIREG STUDY

This i’s not primarily a quantitative assessmentof anticipated marketshares for BEVs,
but rather a detailed examinationof the potential pivotal variables in the BEVmarketand of the
lift, style planningthat householdsare likely to employin evaluating BEVsfor their ownuse.
Wereport frequencies and somemeansand deviations, but we caution against generalizing
the, se numbers,even to the population from whichPIREGinterviews are sampled. Theresults
presented here are themselveshypothesesto be tested on a larger scale in the next phase of our
research. These results do suggest, however, that previous BEVmarket studies mayhave
serious validity problems.

Second car or hybrid household?

Previous BEVresearch has improperlyframedthe decision context for electric vehicles
in terms of first, secondand even third cars. Properly framed, the purchaseof BEVsis a
choice betweena homogeneously-fueled fleet or a hybrid fleet. This is a newconcept for
householdsand requires interaction betweenthe researchers, whohave technical knowledgeof
the vehicles, and households, whoultimately knowwhat such vehicles meanfor their lifestyle.
Travel behavior studies have often conceptualized BEVsas a "second" car, using one
or moreof the followingcriteria: order of purchase;rankingof value; amountof use; or priority
of use comparedto the other vehicles in the household.In fact, the use of the term "second
car" is a cultural-historical concept, related to the perceivedpurchasepatterns of the nuclear
family in the post-worldwar period, in whichyoungfamilies first acquired a car whichthe
father used for workand a second car was purchasedfor mornto do shopping. Onecan still
read this anachronisticcultural interpretation in manydiscussionsof the role of electric vehicles
(e.g. DeLorenzo,1993), despite radical shifts in vehicle buying habits of householdswhich
include the specialization of vehicles and the increased role of women in automotivepurchases.
Someargue that the newest car in most householdswouldnormally be the preferred
long rangevehicle, and that this vintage and use pattern defines the "first" vehicle. But in our
sampleof households, the issue of which vehicle wouldbe the long range vehicle was
determinedby vehicle bodystyle, not order of purchaseor vehicle age. Universally,
householdsin the PIREGsampleuse their larger vehicles, such as minivansor larger sedans,
for long range travel becauseof their comfortand payloadcapacities. Thesmaller vehicles,
whichin the PIREG sampleare also the newestvehicles, are used primarily for local driving,
yet travel nearly as far as the larger vehicle in the household-- an averageof 34 miles per day
for the smaller, newervehicle versus 37 miles per day for the larger, older vehicle. Thusin
these householdsthe "secondcar" label does not accurately describe any vehicle in the
household.Wefind the terms "first" car, "second"car, etc. to be bereft of analytical content.
Therefore, wedo not assign the BEVa "second"car role in the household, but allow the
householdto explore the waysin whichthe BEVchanges their assignmentof vehicles to
drivers and tasks.
Drivers are often, buy not always, assigned to a specific householdvehicle. The
exclusivity of this assignment varies greatly betweenhouseholds. Somehouseholds have very
exclusive his-hers vehicle purchaseand use patterns while others havea pattern of daily or
eventrip-by-trip allocation of vehicles to tasks and drivers. Therefore,at least one decisionstep
for householdswhenconsideringa BEVis to first estimate the routine trip allocations expected
for the vehicle being purchased,as well as special, occasional demandson the vehicle. This
decision involves weighingthe adaptations required for those occasionaltrips against the
benefits of BEVs,such as homerecharging or low fuel cost.
In the PIREGinterview, wet-u-st ask householdsto imaginereplacing one of their
vehicles with a BEV(note that this is not a purchaseintention, but a use intention). Since
originally recruited householdsbased upontheir havingrecently purchaseda car of the likely
bodystyles of BEVs,we call this newlypurchasedvehicle the target ear -- the vehicle we
hypothesizewill be replaced by an electric. However,not all householdsselect the target
vehicle to be a BEV:7 choose another vehicle in the household. Additionally, 3 households,

7
with the incomeand car purchasehabits to back it up, stated they wouldlikely add an electric
vehicle to their householdstock of vehicles and 4 householdsfelt incapable of using a limited
range vehicle in place of any of their vehicles. Thus, 37 householdschose the target vehicle.
Oncethe householdchooseswhich vehicle will be electric, we proceed with the
interviewon the premisethat the electric vehicle is exactlylike the gasolinevehicle it replaces,
except for its driving range and recharge characteristics. Householdsmust next discover the
minimumrange to which they could imagine adjusting.

Minimum range
After examiningtheir travel activity recordedin their 7-daytrip diaries, households
were asked to decide upon a minimum range to which they could adapt without undue (in their
view) sacrifice of their lifestyle goals (whichthey define). Thehouseholdis asked to explain
howthey arrived at such a range, and asked to relive the previously-recordedweekusing their
hypothetical hybrid fleet of vehicles. Theinterviewer interjects with problemsbased on the
knowncar use patterns of the householdand potential emergenciesin order to challenge the
selected minimum range. The point is to explore howthe household adapts to these unplanned
activities -- whetherthey adjust their minimum range requirement, reassign the household’s
vehicles, or makeother traveI adjustments. Thefirst columnof Table 1 reports the minimum
range selected by households.
A substantial numberof householdscan adapt to even very short ranges. Fourteen of
the 51 householdsadapted to a 40 mile range vehicle and I2 moreto a range of 50 miles. A
total of forty-seven householdsadapted to a range of 120 miles or less. Onlyfour households
simplycould not accept a range limit without also havingfast chargingcapability; the vehicles
either were not parked during the day wherethey could be recharged at a slow or "normal"rate
or they were always on the move.

Whatdo you do whenyou need to go farther?

Theinteractive-problemsolving quickly reveals it is too simplistic to assumethat the
BEVsimply substitutes for one of the household’sgasoline vehicles. Substitution is only the
starting point for the household’slearning. Householdsquickly recognize the possibilities for
reallocating travel betweenvehicles and drivers as well as other strategies for incorporatingthe
BEVinto the householdfleet.
Belowwe describe each of the strategies used by householdsin incorporating the
hypothetical BEVinto their fleet of vehicles° Wealso report the numberof householdswho
suggested they wouldemployeach strategy.

1. Workrecharging: Weallowed persons whoworkedat locations with I00 or more

employeesto assumetheir employerswouldoffer recharging facilities. Cars whichare parked
for several hours at workare capable of effectively doublingdaily range. This is the second
2most frequently chosen adaptation to specific hypothetical range problems(22 households).

2. ~ Whenthe driver who normally uses the BEVneeds greater range on a

given day, he or she swapsvehicles with another driver whoserange needs are less on that
day. This wasthe most frequently chosen adaptation (25 households)in the face of a range
problem.This behavior is already in the repertoire of manyhouseholdsin the sample. They
currently swapvehicles for severn reasons, e.g. increased passenger or cargo capacity for a
day° Conversely, a few households never swap. These include households in which drivers
identify strongly with their vehicles, or in whichone driver does not like driving other vehicles
becauseof standard transmissions, vehicle size differences, passenger load needs, or special
cargo requirementssuch as child seats or business supplies whichare inconvenient to transfer
betweenvehicles.

3. Switching: Whenthe driver of the electric vehicle needs morerange, madthere is

another, unusedvehicle in the household,that driver either switches to the unusedvehicle at

8
the,. beginningof the day in anticipation of a needfor additional rangeor returns hometo switch
velNcleslater. This is the third mostfrequently chosenadaptivestrategy, used by 21
households to solve a range problem. This strategy is most commonlyemployedon weekends
whenboth drivers traveled together and one car wasidle.

4o Daytimechareing: A BEVuser mayrecharge during the day at homebetweentrips in

order to extend daily range. This strategy is employedby 10 householdsto solve a specific
problem,but additionally is expectedas a routine practice by 16 households,often on
weekends.If homeserves as the hub of daily activity, as in the case of retired personsand
home-workers,daytimecharging at homeis morelikely to be an important strategy.

5. ~: Wedefined fast charging as being available at a special fast charging

station whichoffers up to 80%of full charge in 20 minutes. Only4 householdsexpected to use
this strategy to solve any range problem.Fast chargingwaswidelyrejected as a solution
becauseof two issues: the long in-station time of twentyminutesand households’desire to use
the gasoline vehicle for any travel extendingbeyondtheir routine activity space. Drivers who
indicated they wouldlike to fast chargehaveno time during the day to park their car at a
possible slow charging location.

6. Carpooling/vanpooling: A household overcomesa range problemby either traveling

together in a householdgasoline vehicle, or traveling in a gasoline vehicle not ownedby the
household. Note this definition extends well beyondtrips to work. For example, two couples
traveling together to a social activity constitutes a carpoolunder our definition. The4
householdsusing this solution live in OrangeCountywhererecent legislation has created
employer-providedcarpool incentives and heightened household awarenessof this option.

7. Renting: A householdor individual driver rents a gasoline vehicle to overcomea

range problem. Onlyone household suggested renting to solve a specific problem. Households
often describe vehicle renting as too expensiveand too inconvenient.

8. Borrowing:A household borrows a gasoline vehicle from another household to

solve a range problem. In 4 households, borrowingthe car of a friend or family memberwas
suggested as a reasonable adaptation, especially in emergencies. Borrowingwas a normal
practice of these households°

9. Bike, walk, transit: Ahouseholduses a bike, walksor uses transit to solve a range
problem. Twohouseholdssuggested biking and 4 suggested transit to solve specific problems.

10. Chauffeur."Gasoline vehicle driver in a householdchauffeurs the BEVuser to a

destination which wouldhave strained the limits of the BEV’srange. Usedby 2 households,
in both cases the householdwas madeup of retired persons.

11. Rescheduletrip_s: Householdor BEVuser reschedules trips from a "long range"

day to a day on whichfewer or shorter trips are made.Reschedulingis used to solve range
problemsby only 2 householdsin the minimum range and adaptation phases of the interview,
but is used by 5 householdsto optimize BEVuse during the final interview phase.

12. Cancel~reduce trips: Thehouseholdor the BEVdriver cancels or shortens a trip to

solve a range problem. Trips were only canceled or reduced in 4 householdsand then only in
the last part of the interview as an optimizingaction.

Transition from minima| to comfortable driving range

After determining minimum
acceptable ranges, exploring adaptive strategies, and
determining howmuchof the householdstravel could be shifted to a hypothetical BEV(which

9
has the minimumacceptable range), households were given the opportunity to "buy" a BEV
with a different driving range and recharging time. Thegoals of this section were: to see
whether households wouldchoose to buy longer driving range than their minimum range; to
observe trade-offs betweenrange and recharge time; and to explore responses to different
electric vehicle technologies(e.g. fast recharging and fuel cells) whichoffered more
"gasoline"-like performance(i.e. longer rangeand faster refueling at fuel stations).
Driving range and recharging times were bundledtogether in "option" packages and
presented in a priority evaluator (PE) table in whicheach range/recharging option wasassigned
a relative price -- that price beingthe difference betweeneach range-rechargeoption and their
minimum adapted range (their "base" option). Thus, these relative prices were customizedfor
each household. In general, householdscould chose to pay a higher price than the base option
by buying longer range or faster recharging, or to pay a lower price by buyingshorter range or
slower recharging. Someof the range-recharge options have a zero price difference from the
base option bundleso that certain levels of range and recharge rates can be chosenwith no
hypothetical dollar cost (or savings). Wedo not report the prices used becausethey are not
indicative of either willingness-to-payfor rangeand rechargingtime or the cost of actually
providingspecific range and recharge times. Consistent with the goals for this phase of the
interview, the prices simply serve to stimulate further householdexploration of range and
recharging. Wereport on recharging rate choices later in the paper. Thesecond columnof
Table 1 reports the ranges chosen by householdsin this phase as "comfortable range" and
Table 2 showsthe transition matrix for changesfrom the minimum range to this comfortable
range.
Table 2 indicates that no householdschose to "buy" a range shorter than their minimum
range. While34 householdscould adapt to a minimum range of 60 miles or less, only 8
householdschose those short ranges whenoffered the range-recharge bundles in the PE table.
Twelvehouseholds chose a 100 mile range. One hundred appears to be a "magic" number-- a
round numberwhich perhaps represents a perceived threshold required to accomplishlifestyle
objectives. Four households whoalready had high minimumranges of I00 to 120 miles
movedto an even higher range of 150 miles.
Theserange choices are labeled "comfortable"becauseit allowed themto either
eliminate all adaptive behaviors previously necessaryto adapt to their minimum acceptable
range, or to reduce the frequencyof such adaptations to only a few occasions per year. In the
end, three-fourths of our householdsdiscovered that replacing one of their vehicles with a BEV
with 100 miles or less driving range madevirtually no difference in their ability to accomplish
their chosenactivities.
Weobserved tremendousinstability in households’ choices of range whenpresented
with incremental information and newproblems. Reflection on their owntravel and the
introduction of information on BEVrange and recharging elicited various range and recharge
rate selections at different points in the interview. Householdsoften discussed their selections
at length and it wasnot uncommon for householdsto makethree different range choices during
the minimalrange, adaptation and comfortablerange stages of the interview.

Newhousehold decision variables related to driving range

Weobserved three critical newdecision variables in households’decision making
regarding limited range. These variables have not previously appeared in survey workon the
BEVmarket. Thefirst is driving-range safety buffers, the secondis routine activity space, and
the third wecall the critical destination.
Thesafety buffer wasthe rangeto be left on the electric vehicle at all times. (As
feature of the BEVweposited instrumentation whichprecisely estimated the remainingdriving
range at all times.) Mosthouseholdsin this study respondedthat they wantedtwenty miles of
range alwaysleft on the vehicle. Householdsarrived at this 20 mile buffer almost irrespective
of the types of hypothetical emergencysituations created by the interviewers or the location of
the householdwith respect to emergencyservices and activity locations. Fourteen households
wereable to adapt to a 40 mile range vehicle whilemaintainingthis safety buffer - that is, they
regularly accomplished their daily activities whiletraveling less than 20 miles in one of their
vehicles.
Routineactivity space contains the locations of activities that the householdaccesses on
a routine basis, including work,schools, doctors, banks, favorite shoppinglocations, grocery
stores, local family and friends. Householdsare quite awareof this set of destinations and
mostare familiar with the distances to these activities. Theseroutine activities formthe basis
for the initial judgmentof driving range needs and the value of homerecharging. For
householdswith geographically small routine activity spaces, homerecharging provides
virtually all rechargingneedsand is therefore quite attractive.
Thecritical destination is the furthest destination whichthe householdmember using
the BEVfeels they must be able to reach, perhaps for an emergencyor for activities important
to their lifestyle. Theymustbe able to reach this destination evenwhenthe gasoline vehicle is
not available. Common critical destinations are a friend or family member whois visited
frequently or a favorite recreational location. Not all family members choosethe samecritical
destination. In somecases, one member identifies a destination whichis already part of the
household’sactivity space, either within the routine activity space or otherwisefrequented.
Anothermember mayargue for an imaginarycritical destination, a destination whichis not
actually in the travel history of the household,but whichthe person nevertheless wants
available.
Sometimessingly, sometimesin combination,these three decision variables act to
deteJ.axtine the driving range choicesof householdsin our study. Whatis clear to us is that
householdshave widely varying degrees of self-awareness of these elements prior to actually
documenting their travel in their travel diaries. Facedsuddenlywith a distance budget(in the
form of a BEVpurchasedecision), it appears most householdswill require a period of learning
to determinetheir range choices.
It is true that in somehouseholds, one memberdid express discomfort with buying a
vehicle with any range limits. Their objections were not based uponcalculation of
demonstratedneeds, but rather upondesires for the open-endedpossibilities of long range and
fast refueling. However,in the midst of lifestyle planningwith their partner during the
interview, they agreed that the householdwas not greatly inconveniencedby having one limited
range vehicle. Only in households in which each driver madehighly autonomousauto purchase
and use decisions did desires for unlimited range prevail over the practical reality of howand
wherethe householdactually travels°

Response to long recharge times

Facedwith "refueling" times measuredin hours, householdswere far moresensitive to
changesha driving range than recharging time. Thebase range-rechargeoption in the PEtable
for each householdhad a recharge time of either 4 or 6 hours, dependingon the range. Most
householdsthen chose 6 or 8 hour recharging. That is, householdsalmost universally
preferred to solve any problemsby increasing the BEVsdriving range, rather than decreasing
charging time. Thesingle most common choice pattern in the PEtable wasfor householdsto
find a longer range, but slower charging, vehicle which"cost" the sameas, or only slightly
morethan, the base option vehicle. Only3 householdschose the fast charging option (at any
range). Wedescribed fast charging as taking twentyminutesfrom arrival at the station to
departure to get 80 percent of a full charge. Mostparticipants could not imaginespendingmore
time in stations than about ten minutes-- twenty minutessimply wasnot "fast". The few
householdsinterested in fast charging also wantedlonger ranges either becausethey intended
to use BEVsfor long distance driving or their vehicle is not parkedat workor at homefor any
substantial time during the day and thus cannot be conventionallycharged during the day.
Rechargingof BEVsis an unfamiliar concept, and muchof the interview is spent
discussing howrecharging works. Manysurveys and reports on BEVstell readers that a BEV
with 100 miles of rangetakes, for example,six to eight hours to recharge. That technical fact
explains a certain limit on BEVs,but does little to help potential BEVusers imaginewhatthey
woulddo on a daily basis or what sort of problemsthey might encounter. Our PIREG
householdsdiscovered there are only a few days per year they wouldactually drain the
batteries lowenoughin their con~eortable-range vehicle to require a full six or eight hour
recharge. Wefind in our interviews that most daily recharging wouldentail only 2-3 hours
recharging at homeat night to recover 20-50 miles of daily use, plus occasional daytime
recharging at workand other daytime parking locations. This daytime charging maybe
especially valuedfor lead-acid batteries, whichperformbetter and last longer if not deeply
discharged.
Thesheer quantity and complexityof informationregarding recharging characteristics is a
barrier in the BEVmarket. Rechargingrates for manybattery types are not constant -- e.g. for
lead-acid batteries, energyis recoveredmorequickly in the fh’st portion of the recharge period.
There maybe manytypes of batteries in different BEVsand manytypes of recharging outlets
with distinct costs dependingon location, time-of-dayand size of electrical service. All this
makesBEVrecharging sound difficult, complicated and inconvenient. But in households
whoselifestyles fit within the distance budgets of BEVs,recharging maynot be an
inconvenience; just the opposite, homerecharging and workrecharging can eliminate gasoline
station visits. Dependingon trip patterns and time demandson drivers, as well as attitudes
towards gasoline stations, homerecharging is a substantial perceived benefit for many
households. Manywomenwhoare the primary caregivers to the family’s children stated they
did not like to go to the gas station with the children in the car. This necessitated schedulinggas
station visits betweentrips in whichthey were chauffeuringchildren, or requiring their spouse
to refuel the vehicle. Short, home-basedtrips whosesole purpose is to buy gasoline were not
uncommonin our sample.

Flexibility of vehicle assignments to household trips

Thedistance traveled in the vehicle replaced by a hypothetical BEVchangesvery little
in the search for the minimum range to whichthe householdcan adapt. In essence, the
householdsfind the minimum range BEVthey can use to accomplishthe sameactivities in their
diary week. Onlysix householdssubstantially reduce (>10%reduction) the proportion of the
householdtravel assigned to the vehicle selected to be replaced by the hypothetical BEVo
Thirty-two householdsmakeno changesin the use of the selected vehicle°
In the last phase of PIREGinterviews, householdsoptimized the use of their
hypothetical hybrid fleet under the premisethat gasoline prices had risen to $5 per gallon. The
hybrid fleet consists of a BEVof the minimum-adapted range and all the household’s other
ICEVs.Theintent of this exercise wasnot to test their responseto expensivegasoline per se,
but rather to test their ability to reallocate driving betweenvehicles under conditionswhich
greatly favored the use of one fuel (electricity) over the other (gasoline). Whenpressed
high gasoline prices to increase the use of their BEV,virtually all the householdsshifted
additional travel to it. Three householdsthat did not are poor candidates for BEVsbecause
they had greatly reduced the miles assigned to the BEVin their earlier search for a minimum
adapted range and wereunable to shift travel back to the BEVin the optLrnjzingscenario. After
optimizing vehicle use, the BEVsin the 48 other householdstraveled an average of 17 percent
moremiles than the gasoline vehicle it replaced had traveled during the diary week. Half the
householdsshifted at least12 percent moretravel to the BEV,and one-fourth of the households
shifted at least 25 percent more.
Theseresults further support the idea that analyses whichassumesimple substitution of
a BEVfor one householdvehicle are too simplistic. Undera variety of conditions, households
are able to shift travel betweenhouseholdvehicles. Ourinterview uses a "shock"technique to
force householdsto use fuel prices to makethese adjustments. But the results suggests a
dynamicin whichnewhybrid householdsmayshift increasing amountsof their travel to a
BEV,if they realize the promiseof lower operation and maintenancecosts or are offered
vehicle use incentives such as preferred parking and use of tfigh-occupancy-vehicle (HOV)
lanes.
Market Segments: Household’s Adaptability to BEVs
Basedon the preceding results, weclassify householdsinto one of three categories
with respect to their ability to incorporate a BEVinto their householdfleet. Wefind 29 of the
51 households are pre-adaptedto BEVs.To use a truly limited-range BEV,these households
require no changein their travel behavior, no away-from-home recharging location (with the
possible exception of workplace recharging in 13 cases), and no changein the bodystyles of
ve, hicles they buy. Their routine activity spaceis small in geographicalextent andtheir critical
de, stination is easily reachedin a BEV.The"comfortable"rangeselected by all these
householdsis less than 103 miles -- sometimesmuchless.
Weclassify another group of 15 households as easily adapted. These households
infrequently switch vehicles or swapvehicles betweendrivers to accommodate a limited range
vehicle. Thesebehaviorsare already part of these households’repertoire of travel behaviors.
Theyare morelikely than pre-adapted householdsto want to recharge at workduring the day.
Theirroutine activity spacesare larger in size and their critical destinations are farther away.
These households choose "comfortable" driving ranges between 80 and 103 miles.
Lastly, 7 households are non-adapted.These householdsexhibit one or morebehaviors
wlfich makeit difficult or impossiblefor themto include a BEVin their householdstock of
vehicles: they havelarge routine activity spaces resulting fromlong or uncertaindaily trips for
work(as opposedto trips to work), a critical destination beyond150 miles from home,
inl~exibility in the assignmentof householdvehicles. Absentlarge changesin their routines,
lifiestyle goals or budgets, these householdswouldnot purchasea BEV.

WHY DO PREVIOUS STUDIES ESTIMATE HIGH PENALTIES FOR

LIMITED RANGE?

Oneinescapable conclusion of the PIREGinterviews is that manymulti-car households

easily adapt to driving range limits on one householdvehicle. Howthen do weinterpret
previous research whichestimates extremelyhigh disutilities for limited driving range? Oneof
our hypothesesis that householdsdo not understand, or not awareof, the distances of either
the, ir individual trips or daily travel. To understandhowwell householdsknowtheir travel
distances, we asked PIREG participants to estimate each day’s travel at the start of the day and
each trip’s distance at the start of the trip. Theyalso recordedodometerreadings, which
allowedus to calculate the actual distance traveled each day and each trip.
As expected, participants were muchbetter at estimating travel distances for regular
trips than irregular trips. Commute trips are the most common type of regular trips. Commute
trip length in our sampleranged from I to 45 miles, with a meanof 12.6. Thedistance of the
first recorded commute trip wasgenerally estimated accurately, with an average absolute error
of 1.1 miles, or 8.7%of the meanestimate. Greater absolute and percent errors were madein
estimating the distance of the longest (non-commute) trip madeduring the week, which
typically a trip to an irregular destination. Nomatter howthe data are analyzedto distinguish
betweenerrors due to irregularity of the trip and errors due to the length of the trip, wefind
ave, rage errors aroundthe meanestimate that are 3 times larger than the commute trip errors.
Analyzingonly those longest non-commute trips that are less than 45 miles in length to remove
the possibility that peopleare simplyless able to estimate longer distances than shorter
distances, yields a meantrip length of 22.5 miles and a meanerror of 5.0 miles (24%).
also selected the worstestimate for each driver, regardless of the trip type or length. The
ave, rage worst error was34%of the actual trip distance. Since initial commute distance
estimates have an averageerror of only 8.7%,drivers are clearly estimating other trip distances
with muchless accuracy.
Drivers showedimprovementduring the diary weekin estimating trip lengths. The
average error in estimatedtrip distance betweenthe first and third recordedcommute trip
droppedfrom 8.7%to 1.2%0The act of estimating and measuringtrip distances was itself a
learning process for manyhouseholds. Several householdsindicated the diary had increased
their awarenessof both the numberof trips they madeand howfar they traveled.
Onthe issue of total daily distance estimates, weexaminedthe most poorly estimated
day for each driver. Drivers had moredifficulty estimating long travel days, just as they had
difficulty with irregular trips. Theaverage absolute worst error in estimating howfar an
individual driver wouldtravel on a given day was17 miles or 34%of the average actual
distance for that day. Householdswhich underestimatedtheir longest day tended to do so by a
larger amountthan householdswhich overestimated their longest day; the meanunderestimate
was 28 miles and the meanoverestimate was 18.6 miles.
Despitethe magnitudeof these errors, wecannot attribute the high disutility of limited
range inferred from stated preference to these errors. Wecreated worst case scenarios by
adding a householdsworst absolute error to its longest travel day (regardless of whetherthey
actually occurred on the sameday). Thesumis still muchless than the driving range of
gasoline vehicles and well within the range of the BEVsconsideredin this study for all but a
few households. In only four householdsdoes the sumof the longest clay and the worst error
exceed 150 miles. Note wehave addedthe absolute value of the worst error, thus inflating the
apparent range requirement.
It is not the travel estimation errors per se whichare the source of the apparent
disutilities of limited range vehicles, rather it is consumerschoice of a familiar anchorwhen
faced with the unfamiliar prospect of a distance budget. Althoughit wasnot a formal part of
the studys, manyPIREGparticipants stated at the end of the interview that prior to completing
their travel diary, they wouldhave offered muchhigher estimates of their daily minimum range
needs. But after reviewingtheir actual driving needs and lifestyle goals, they foundthat they
could adapt to a limited ranged vehicle. Householdsdeal daily with moneyand time budgets,
but have no daily distance budgets, and currently do no accountingof distances (unless they do
so for business purposes or to gauge fuel consumptionrates). Whenconsumers, unfamiliar
with a distance budget, are asked to respondto limited ranges, they typically respondwith a
range they are familiar with -- that of their current gasoline vehicles. But those vehicles have
refueling characteristics and range instrumentation optimizedfor open road, long distance
travel. Wehave observedthis anchoringto the familiar gasoline vehicle range in focus groups
(Turrenfine, et al; I992) and believe this behavioris largely responsible for the high apparent
disutilities estimatedby previous stated preferencestudies.

CONCLUSIONS

Our investigation of hybrid househoIdbehavior, motivated by discrepancies between

travel behaviorand stated preference studies and by the conflicting results of attitude surveys,
suggests the following. First, travel behavior studies are flawed methodologically.Their
results are biased not only by their use of one day travel diaries and questions regarding
"typical" rather than actual travel, but moreimportantly by the assumptionthat the BEVis a
substitute for one particular householdvehicle.
Second,stated preferencestudies, whichresult in high disufilities for short driving
ranges, are incorrect. Wefind that hybrid householdsare willing to engagein a wide range of
adaptive behaviorsand that, given the opportunity to reflect on their owntravel, most discover
that a range limit on one householdvehicle is simplynot a problem.Stated preference studies
are likely measuringuncertainty and unfamiliarity, not utility. Thelarge apparent "utility" of
long range maysimply be an artifact of consumerconservatism whenfaced with an unfamiliar
technology.
Third, surveys of "green" attitudes fail to provide useful insights into whomaybuy
BEVs.If we accept the definition of a "green" market as a contingent of consumerswhoare
willing to pay several thousanddollars extra for a zero emissionsvehicle, then wedo not find
strong evidencefor such a market.
 However,we do find evidence for a viable BEVmarket. Webelieve that a large
portion of multi-car householdsare "pre-adapted", or can easily adapt, to the BEVrange and
rechargecharacteristics attainable with technologylikely to be available in 1998, the first year
"zero emission vehicles" must be sold in California. Whatthese householdslack to makea
BEVpurchase decision is information and experience.
Weare not, however,arguing that all the pre-adapted and easily adapted householdsin
this study will purchasea BEV.Wesee significant barriers in the market. In fact wefind little
evidencein this study that most consumersare willing to pay significantly morefor BEVs.We
do find that mostof the householdsin our samplecan easily adapt to vehicles of far shorter
range than previous research suggests and most PIREGhouseholds see a few simple
adaptationsas a reasonablecommitment in the context of an historical shift to clean cars. In
pre-adapted and easily adapted households, BEVswith driving ranges of 60 to 100 miles will
competewith gasoline vehicles at roughlycomparableprices, but only after the requisite
educationand reflection whichconsumerswill require to estimate the value of hybridizing their
vehicle holdings. Ourfindings suggest that the marketfor electric vehicles will require
substantial educational efforts, assisted by governmentand industry.
The problemis that previous market research and thinking about the BEVmarket has
neither explored complexrange and recharging issues with consumersnor provided an
adequate context for consumersto estimate the impact of BEVson their lives. Nor has any
previous study adequately addressed the positive utility whichmaybe assigned to home
refi:teling. For this reason, weexpect that previous survey workbased on hypothetical choices
has systematically underestimated the BEVmarket because consumersare unfamiliar with the
potential waysof combininggasoline and battery electric vehicles in their households°Future
survey workcould be improvedif it incorporates information on the adaptive behaviors and
critical decision variables used by our PIREGhouseholds: vehicle switching and swapping,
homeand workplacerecharging, range buffers, routine activity space and cdtical destinations.
Additionally, it is prematureto focus on consumerpreferences for BEVs.Our work
demonstratesthat preferences for driving range and recharging are as yet unformed.In a
complexand rapidly changinginformation milieu such as that aroundelectric vehicles, weare
well advisedto create experimentaland experiential contexts. Givenall this, marketresearch
must focus less on providing questionable marketpenetration estimates and instead workto
identify potential marketsegmentsfor BEVsand to understandthe nature of market barriers.
Oursocial goals of energyflexibility and clean air and our private goal of accessibility to
desired activities will be better served by an understandingof marketdynamicsand information
on howto build viable marketsfor electric transportation options than by contentious and
unreliable estimates of somefuture end-state.

ACKNOWLEDGMENTS

Wewouldlike to thank Martin Lee-Gosselin, Director of the Programme

Interdisciplinaire en Amenagementdu Territoire et Development Regional, Laval University,
Quebecfor his tireless efforts and keen insights during the design and implementationof
PIREGinterviews. This workwas funded by the California Institute for EnergyEfficiency, the
University of California Transportation Center, ExxonUSA,Nissan North America, and
Nissan Researchand Development.We,the authors, are solely responsible for the
conclusions.
REFERENCES

Beggs, S.D. and N.S. Cardell (1980) "Choice of smallest car by multi-vehicle householdsand
the demandfor electric vehicles", Transportation ResearchA, 14A,pp. 380-404
Berkowitz, E., R. Kerin, S. Hartley and W. Rudelius (1992) Marketing. 3rd Edition, Boston,
MA:Richard D. Irwin Inc.
Buist, D.R. (1993) "Anautomotivemanufacturer’s alternative fuels perspective", Proceedings
of the First AnnualWorldCar 2001 Conference, University of California, Riverside:
The Center for Environmental Research and Technology, pp. 51-55
Bunch, D.S., M. Bradley, T.F. Golob, R° KJtamura and G.P. Occhiuzzo (1993) "Demandfor
clean fueled vehicles in California: Adiscrete-choice, stated preference survey",
Transportation Research A, 27A, pp. 237-53
Calfee, J.E. (1985) "Estimating the demandfor electric automobilesusing fully disaggregated
probabilistic choice analysis", Transportation ResearchB, 19B, pp. 287-301
Dables, J. (1992) "Developingthe greatest uncertainty -- the EVmarket", Presentation
ConvergenceNinety-Two:International Congressan Transportation Electronics,
Dearborn, M-I, Oct. 19-21
DeLorenzo,M.(1993) "Plug in, turn on, or drop out: a debate groundedin reality",
Autoweek, 43(51), p.17, December
DeshpandeG. K. (1984) "Developmentof driving schedules for advanced vehicle
assessment", SAETechnical Paper Series No. 840360, Warrendale, PA: SAE
Fairbanks, Maulin and Associates (1993) Reported in "Zapped", Autoweek, 43(50), pg.
December 6
Greene D.L. (1985) "Estimatingdaily vehicle distributions and the implications for limited-
range vehicles"~ Transportation Research B, 19B, pp. 347-58.
Hamilton, W.F. (1983) "A Critical Reviewof Electric Vehicle MarketStudies", Releasable
Memorandum 2446/R1, Santa Barbara, CA: General Research Corporation
Hautzinger, H., et. al° (1991) Elektroauto und Mobilitat -Das Einsatzpotential von
Electroautos, Heilbronn, Germany:Institut fur AngewandteVerkehrsund
Tourismusforshung E.V.
Hu, P.S. and J. Young(1992) Summaryof Travel Trends: 1990 Nationwide Personal
Transportation Survey, Washington,D.C.: Federal HighwayAdministration, Office of
Highway Information Management.
Jones, P. M. (1979) "HATS - A technique for investigating household decisions",
Environment and Planning. A, 11(1)
Kirchman,R. (1993) Report of the Electric Vehicle at-HomeRefueling Survey, Prepared for
Pacific Gas and Electric Co., San RamonCAby Original Research Customer
ManagementServices.
KiselewichS. J. and W.F°Hamilton(1982) "Electrification of householdtravel by electric and
hybrid vehicles", SAETechnical Papers No. 820452. Warrendale, PA: SA.E
Kurani, K.S., T. Turrentine, and DoSperling (1994) "Electric Vehicle Owners:Tests
Assumptionsand Lessons on Future Behavior from 100 Electric Vehicle Ownersin
California"., Forthcomingin the Transportation ResearchRecord
Lee-Gosselin, M. (1990) "The dynamicsof car use patterns under different scenarios:
gamingapproach", In Jones, P. (ed.) Developmentsin dynamicand activity -based
approachesto travel analysis, Aldershot~UK:Avebury,pp. 251-27I.
Morton,A., et al. (1978) "Incentives and acceptanceof electric, hybrid and other alternative
vehicles", Cambridge,MA:Arthur B. Little
Nesbitt, K.A., K.S. Kurani and M.A. DeLuchi (1992) "HomeRecharging and the Household
Electric Vehicle Market: A Constraints Analysis", Transportation ResearchRecord, No.
1366, pp. 11-19
Newsweek(1990) "1990 buyers of new cars: A research report from Newsweek",New
York, N.Y.
J.D. Powers and Associates (1993) The Power Report- 1993 Automotive ConsumerProfile
Study, AgouraHills, CA.: J.D. Powers and Associates
Rutherford, S.G., D.A. Nierneier and C. Kooperberg(i994) "Daily travel patterns and electric
vehicles: Anexploratory data analysis", Forthcomingin the TransportationResearch
Record
Sperling, D. (1994) Furoredrive: electric vehicles and sustainable transportation, Covelo,CA:
Island Press
Turrentine T.S., D. Sperling and K. Kurani(1992) "Marketpotential of electric and natural gas
vehicles", Research Report UCD-ITS-RR-92-8. Davis, CA: Institute of Transportation
Studies, University of California
Turrentine T.S. and D. Sperling (1991) "Theories of newtechnology purchase decisions: The
case of alternative fueled vehicles", Berkeley, CA:University of California
Transportation Center WorkingPaper No. 129

17
Table 1: MinimumAcceptab|e and Comfortable Driving Range
Driving Range Numberof households for Numberof households for
whomthis was the whomthis was the
minimumacceptable range comfortable range
40 miles 14 1
50 miles 12 3
60 miles 8 5
70 miles 3 3
80 miles 6 10
90 miles 0 3
100 miles 2 12
120 miles 2 6
150 miles 0 4
Unlimited miles 4 4
Total 51 51

Table 2: Transition Matrix from Minimumto Comfortable Range

Comfortable range i
Minimum 40 50 60 70 80 90 100 120 ! 150
Range miles miles miles miles miles miles miles miles i miles
40 miles 1 1 1 1 3 1 4 2 i
50 miles 2 3 1 2 3 1"
60 miles 1 4 1 2
70 miles 1 1 1
80 miles 2 3 1
i
90 miles
100 miles " 2
120 miles [ 2
Note: This table does not include the 4 householdswhocould not/wouldnot accept any limits.
FOOTNOTES

1Thevariable of range is separated fromother refuelingkrechargingattributes such as

type of fuel, speed of recharging or refueling. Weselected fromthe data in these studies the 50
mile range to fit the bottomend capabilities of BEVsand the 200mile range to represent the
most extremepossibilities for BEVs,although such a battery system is not yet demonstrated
for conventionalpayloadsand driving cycles.
Zln PIREG wedid not discuss potential differences in on-off peakelectricity rates for
recharging, although such behavioral issues are important for utility load management.We
were most interested in consumerbehavior outside such rates, and it is unlikely consumers
could easily assess their valuation of daytimeopportunitycharginggivendifferent electricity
prices, even in the PIREGinterview context. Ourresults cannot be used to measurepotential
daytimeuse assumingdifferential rates, rather they demonstratehowmuchdaytimecharging
BEVdrivers might do if such charging is easy and inexpensive.

3We do not discuss household’s "preferred" range prior to the PIREGinterview. We

wished to avoid both anchoring to a pre-concieved numberwhich we had observed in previous
workover-dramatizationof the magnitudeof any possible changesin their daily range choices.

19