Projection of Urban Household Automobile Holdings and New Car Purchases by Type

Transcription

1 1 Projetion of Urban Household Automobile Holdings and New Car Purhases by Type CHRSTOPHER L. SARCKS, ANANT D. VYAS, and JAMES A. BUNCH ABSTRACT n this paper a proedure for modeling the hoies made in urban Amerian households among personal vehiles on the bases of ost, passenger apaity, and engine tehnology is disussed, and those preferenes to the years 199 and 2 are projeted. The results of this disaggregate tehnique are used by the other preditive researh tasks undertaken by Argonne National Laboratory in a projet entitled Tehnology Assessment of Produtive Conservation in Urban Transportation (TAPCUT). n these projetions vehiles with standard spark-ignition (Otto-yle) engines ontinue to dominate automobile holdings and new ar purhases in either of two so.ioeonomi senarios under any of three settings (an existing poliy set and two alternative onservation strategies). From 199, small ars (whih seat four or fewer passengers) dominate urban holdings and sales in two of the three TAPCUT energy strategies (the exeption being the strategy that emphasizes individual travel), and this holds true with only a minor variation in both soioeonomi senarios. Advaned-tehnology vehiles are most suessful under the individual travel strategy. Vehile harateristis are far more signifiant than demographi desriptors in estimating household vehile hoie when using this modeling approah. n this paper the method and results of the household vehile hoie model used to foreast the distribution of automobiles bought and used by urban households in eah of two projetion years (199 and 2) are desribed. These foreasts in turn led to estimates of total demand for and ost of travel, and of the future eonomi of automobile prodution areas for the Tehnology Conservation in Urban projet. and environmental impats and operation in urban Assessment of Produtive Transportation (TAPCUT) The TAPCUT projet had the stated goals of providing (a) a desription of several alternative strategies that promote energy onservation in urban passenger transportation, (b) a better understanding of the environmental impats of suh strategies, and () identifiation of the onstraints on the implementation of suh strategies. Two produtive o.nservation strategies were designed to save energy in urban passenger transportation when substituted for poliies urrently in plae. A referene set of impat foreasts was then prepared for these two strategies. One onservation strategy stressed group travel (e.g., transit and arpooling), whereas the other promoted individual travel in private automobiles. The strategies were designed to ause minimal disruption of life-styles and the eonomy while ahieving. redutions in the onsumption of aggregate energy, espeially that derived from petroleum. Travel demand analysis was performed for eah of three typial ities under poliies urrently in plae and foreast to ontinue, and also under the alternative strategies (i.e., group travel strategy and individual travel strategy). Environmental impat analysis of the foreast travel demand under eah strategy was ity speifi and inluded estimation of air and water pollutant loadings along with their assoiated impats on human health. Traffi safety impats were also estimated. Soioeonomi impats aused by vehile use and vehile prodution were assessed. mpats on physial environment, resoures, health, and safety aused by vehile and fuels prodution and infrastruture onstrution were also addressed. The final step was the overall omparison of poliy-driven results to the results obtained under the in-plae poliy set. NTRODUCTON TO MODELNG VEHCLE CHOCES BY HOUSEHOLDS General Approah Disaggregate statistial modeling of vehile hoies by households has a short but stimulating history (_!-j). n his omprehensive review of the topi, Tardiff Cil disusses the requirement that all suh models define household and vehile harateristis so that a suffiient (but not exessive) number of dependent variables are available for estimating oeffiients of a hoie funtion. Speifiation of too many variables may introdue a degrees-of-freedom problem. An appropriate ourse is to establish variables that are interations of vehile desriptors, whih do not vary aross households, and soioeonomi desriptors, whih do vary aross households. Variables may also be established to orrelate the sensitivity of aquiring or holding vehiles by household to employment opportunities and other measures of aessibility to various modes of transportation. n the TAPCUT modeling system household and work plae loations are predetermined by regional population and employment foreasts and the land use poliy for eah senario. Aggregate household travel demand by mode is developed by using an extended version of a sketh-planning transportation model alled extended short-r.ange generalized poliy (XRGP) from household work-trip reords and household harateristis that influene the nonwork travel of its members (see paper by Kaplin, Gur, and Vyas elsewhere in this Reord). Household vehile holdings are part of the reord of eah household's harateristis. To projet total holdings and new ar purhases by household lass it was therefore suffiient to model only the vehile hoie proess of households haraterized for the XRGP model. Only a household-based model of vehile preferenes has an output apable of diretly feeding the rest of the TAPCUT modeling sequene, whih foreasts travel demand and energy use at the level of the standard metropolitan statistial area (SMSA).

2 2 Transportation Researh Reord 952 Foreasts of automobile holdings at higher levels of aggregation than the household (e.g., regional or national metropolitan totals) would fall short of th<> r<>,.olution of the travel-demand and enerqv-use model, whih is household based. Moreover, the TAPCUT projet required personal vehile data (size lass, tehnology, materials of omposition by perentage of total weight, and so on) of the highest possible resolution in order to foeeast the resoure and environmental impats of vehile-related energy poliies. Empirial Considerations The 1977 Nationwide Personal Transportation Study (NPTS) provides a detailed ross-setional data base for the disaggregation of automobile holdings and examination of vehile preferenes (7). This,- household sample also provides onsiderable information that orrelates the demographi harateristis of households (inome lass, number of people, and age and eduation of household head) and automobile ownership by vehile type. The TAPCUT projet used the urban household data from this survey as its base file of national urban soioeonomi data. Vehiles inluded in the survey were aggregated to three size lasses in a household desriptor file before their use in TAPCUT. Unfortunately, the NPTS provides no information on the evolution over time of vehile holdings by speifi types of household. n the absene of onsistent longitudinal data, it was therefore assumed that these holdings are stable through time within a soioeonomi ategory but not within a speifi household, whih an move from one soioeonomi ategory to another. For example, the vehile holdings of low-inome households remain onstant, but an individual low-inome household is assumed to hange its holdings as it moves into the middle-inome ategory. This in turn assumes that demographi or eonomi variables prinipally determine the total holdings of personal vehiles over time. Thus hanges in the demographi struture of the national population, disaggregated to the household level, aount for growth or deline in the total fleet of personal vehiles. These hanges, represented by inreases or dereases in the total households in eah ategory, were estimated for TAPCUT through the tehnique of iterative proportional fitting (PF) (). From this perspetive a given NPTS household type (or ell), whih is lassified aording to ar ownership plus the four demographi variables previously mentioned, will hold as many automobiles in 2 as it does in 1977, although not neessarily the same types of automobiles. This struture permits households of a given type to hange total travel but not total holdings in response to hanges in highway or level of transit servie. The type of vehiles held an hange in response to hanges in automobile operating harateristis suh as fuel eonomy; but, again, the total number held is assumed to remain unhanged. Predition of holdings in year x should be logially onsistent with predition of new ar sales in year x - n (n = 5,1) beause vehile types that do not sueed in the marketplae, in ompetition with alternative types, will not be available as holdings or used ars after 5 or 1 years. Similarly, a projetion of household fleet distribution that shows a vehile virtually disappearing from household fleets in a foreast year points to a essation in sales of that vehile before the foreast year, even though a standard logit probability distribution would guarantee it some fration of sales in that same year. Adjustments to foreast results are required whenever automobile holdings and new ar purhases, whih are projeted independently, do not agree. Prinipal Assumptions n summary the model seleted to projet urban household automobile holdings and new ar purhases had to aommodate two basi assumptions. First, individual households hange their total holdings of personal vehiles only as their respetive demographi status hanges. Seond, the harateristis of the vehiles held are a joint funtion of the household demographi profile speified for a given projetion year and the attrativeness or salability of an available type of personal vehile. MODEL FORM AND PROCEDURE nput-output The disaggregate vehile stok alloation model (DVSAM) was used to estimate vehile holdings for eah of the projetion years. The model uses an inremental logit equation to estimate vehile holding probabilities for existing tehnology vehiles and a simple logit equation to estimate holding probabilities for new tehnology vehiles. t is derived from a vehile hoie model developed by Lave and Train (,_2) The prinipal purpose of using an inremental analysis, as opposed to running the Lave-Train model diretly, is to evolve vehile-holdings patterns from predetermined onditions at a given window in time, thus avoiding the time and ost of generating annual new ar sales estimates and applying a umulative srappage funtion. This is aomplished by inorporating given information on the dis_tribution of holdings (in this ase, the 57 x NPTS array) and by using the model mainly to predit hanges in this distribution. A vehile-holdings ffre that uses 1977 NPTS data was reated for 19. For all subsequent projetion years, the output vehile sales file from the DVSAM run for the preeding foreast year was used as the base year file. To maintain onsisteny with the vehile total treated in the Lave-Train hoie set and to avoid diluting future probability distributions through the introdution of spurious hoies, the maximum permitted size of any vehile type hoie set is 1 in DVSAM. Given a hange in the utility of automobile type a i, for example, to household type h (as omputed by the Lave-Train funtion), and given the base probability of this utility resulting in a new ar sale, the predited probability of the holding an be expressed as follows: P 'h(a;) = exp[liui P (a;)]/ { exp[liuj P (aj)] } where A hoie set of vehile types available in both base and foreast years, all ompeting vehile types in the set, hange in utility of vehile type ai to household h, and base sales probability for vehile type ai. Beause this hoie set expands in eah suessive projetion year (from vehiles in 19 to in 199 and to either 9 or 1 in 2), it is neessary to partition the logit proedure to predit the share of newly introdued vehiles aording to their omputed absolute utility. Shares of all (1)

3 Sariks et al. 3 vehiles in a given year are first d1stributed aording to a standard logi t formula (by using absolute utilities); then the normalize perentage share of all vehiles that appear for the first time in that year's hoie set (i.e., new tehnology vehiles) is subtrated from one. The remaining perentage represents the aggregate share for onventional, or established, vehiles. This share is distributed among those vehiles by inremental logit aording to the omputed hange in the utility funtion from base to foreast year, as shown in Equation l. Thus for all but one of the strategy and senario ombinations, six established and two new vehile tehnologies are defined for 199, and eight established tehnologies and one new one are defined for 2. The exeption is senario under the individual travel strategy, in whih two new tehnol ogy vehiles are defi ned for the year 2. n eah ase standard and inremental logi t formulas are applied sequentially. Figure l shows the data file flow of the DVSAM. brated model must be onsidered unaeptable for purposes of onsistent foreasting. Therefore, the Lave-Train funtion was alibrated away from 1977 equilibrium by systematially modifying the dummy variable oeffiients for all Otto vehiles in order to- repliate available information on 19 sales aod holdings distributions that use the logit funtion. This was aomplished through use of a reursive tehnique to alter the utility values. n this tehnique the upper or lower bounds' on the perentage share of holdings are speified for any vehile type (s) with whih a dummy variable is assoiated; then the urrent value of the share is multiplied by the log ratio of the nearest desired bound and the urrent value i subseq uently, the utility omputation and logit proedure iterat es to losure. PROJECTED HOUSEHOLD AUTOMOBLE FLEET HOLDNGS AND NEW CAR SALES BY SCENARO Two eonomi and soial-organization senarios were defined for the TAPCUT projet; they differed in SCENAR JO/POL CY PARAMETERS BASEL! NE SALES DSTRBUTO OUSEHOLD AUTO M HOLDNGS RATES (1977) T D NPTS HOUSEHOLD COUNTS PF - GENERATED HOUSEHOLD COUNTS VEH!CLE CHARACTERSTCS F LE K OUTPUT PROB L TY DSTRBUTON Arm SUMMARY TOTALS FACTORED tlh! CAR PURCHASES XSRGP VEHCLE HOLD NGS THROUGH CTY PFs FGURE 1 Data file flow in DVSAM. - --Sales Projetions Only Adj us tmen.ts A pronouned and inesapable result of the first stok estimation--developing a 19 distribution from the empi rial 1977 values--was its failure t;o repliate the dramati movement to smaller ars sine 197. n the 1977 NPTS holdings the onsiderable bias t owa r d large ars (the urb weights of more than 3 perent of automobiles and vans exeed 3, 7 lb) reates a n i nertia i n t he log i t dis tribut i o n that imposes past resul t s on futur e p roj et i ons. This is part i ularl y s t rong when, as i n the TAPCUT senario indivi d ual travel poliy setting, mileage and performane improve at a onsistent pa e aross all vehile types. ndeed, as holdings distributions inremented only on 1977 holdings are modeled beyond 19, the trend in automobile aquisitions and holdi ngs by size tend.in a dire l:lu n exatly opposi te to what has been observed over the past 3 years. Even if the reent move toward smaller veh.ile s i s held t o be s ho r t t e rm a nd eventually reversi ble (wh ih does not appear entirely reasonable g iven expeted ha nges i n household omposition and size), this anomalous behavior of the unaligross national produt (GNP) growth rate, soial organization, retail fuel prie, total metropoli tan population, average household inome, environmental _regulations, and types of fuel available for transportation. The two senarios an be briefly distinguished as senario --a wealthy eonomy with high tehnologial suess--and senario --a relatively poor eonomy with low tehnologial suess. National urban and ity-speifi foreasts of population and employment harateristis were prepared under eah senario. Additional data regarding these senarios may be found in LaBelle et al. (). Prototype ities were seleted by using a fatoranalysis tehnique that identified extreme ities along three dimensions relevant to transportation energy use (1). One dimension, alled Megatown, identifies large ities with satisfatory transit s yotms. The seond dimension, Sprawlburg, typities newer, fast-growing, sprawl ities. The Slowtown dimension identifies midwestern industrial ities that are smaller in population than the other two. metropolitan areas in the nation were related to these three dimensionsi an expansion method was then developed to make national urban foreasts based on

4 Transportation Researh Reord 952 the detailed foreasts of the three typial ities seleted to represent the three dimensions. Automobile and transit vehile harateristis """'&.'- prjted in detail undr,,.,ral '?t:!=i nf poliy and senario onditions. Three different sets of vehiles were used in the analysis: set, the expeted tehnologies, was used for the in-plae poliy and group travel strategy in both senarios 1 set A, designed as the best tehnology for both onservation and performane, was tested for the individual travel strategy in the optimisti senario: and the third set, a modifiation of set, was tel!ted in the other senario under the individual travel strategy. Vehiles were haraterized by size lass, engine type, fuel eonomy, emissions profile, purhase prie, operating osts, materials omposition, and (for personal vehiles) performane. Household Fleet Holdings in a 'l')'eial City Figures 2-7 show yea-2 distributions of automobile holdings aross several vehile and demographi harateristis in the SMSA of the TAPCUT ity Spawlburg. Sprawlburg values are given here beause this ity type is the most influential in the TAPCUT expansion proedure for generating a national profile of urban travel effets. Figures 2 and 3 show the distribution of the at:tmbilq flet in Sprawlburg by :ehile tehnology under eah poliy or strategy in senarios and, respetively. The tehnology mix is nearly idential under the in-plae and group travel settings in both senarios, but there are signifiant tehnologial differenes between. vehiles in the individual travel strategy in senario and those in senario. n senario vehiles are uniformly superior in eah tehnology to those available unde.r the in-plae poliy, and ars with reiproating external-ombustion (Stirling-yle) engines are not available under any other poliy. Not surprisingly, beause of idential vehile harateristis, the perentage distribution aross tehnologies is similar between senarios for eah of the first two strategies. Vehiles with standard spark-ignition (Otto-yle) engines dominate all distributions, with diesels ahieving their highest penetration (15.5 perent) under the individual travel strategy in Senario. Penetration by eletri ars reahes only 2.3 perent under the group travel strategy in both senarios, a performane attributable to relatively high operatin g ost (whih inludes replaement of battery paks) and the range limitations of eletri ars. Only under the group travel strategy (with high petroleum fuel taxes) does the life-yle ost of eletri ars beome ompetitive with that of heat-engine ars on a per-mile basis. 1 V) O>,!::; "U 2 i Stlrllng. Zl Eletrls D Dlesel, l'z3 Otto, Brayton n-plae Group ndividual FGURE 2 Fleet distribution by vehile tehnology in Sprawlburg by strategy in senario in 2. 1 V) O> ' 2 n-plae Group ndividual St!r!!ng Brayton Zl Eletris CJ Diesel l'z3 Otto FGURE 3 Fleet distribution by vehile tehnology in Sprawlburg by strategy in senario in 2. Figures and 5 show tile distribution of the Sprawlburg household fleet by vehile size under eah energy strategy in senarios and, respetively. From 199, small ars dominate urban holdings and sales under the in-plae poliy and the group travel strategy in both TAPCUT senarios, and this phenomenon is manifested in the year-2 fleet distribution. The trend to these four-passenger ars is damped under the individual travel strategy by improved mileage and performane in every vehile of the hoie set. V) O> ' V) O> 1 2 n-plae Group ndividual FGURE Fleet distribution by vehile size in Sprawlburg by strategy in senario in n-plae Group ndividual FGURE 5 Fleet distribution by vehile size in Sprawlburg by strategy in senario in 2. Zl Largo D Medium l'z3 Small ZJ Large D Medium Z(! Small Fi9ures and 7 have been inluded to emphasize that senario variables--in this ase inome distribution--do not vary by poliy. The number of house-

5 Sariks et al. 5 holds alloated to eah inome group and the number of automobiles held by eah are fixed throughout the senario. This means that although the ownership patte rn by size and tehnology varies aross poliy settings, the total holdings, and thus the perentage of vehiles held by eah inome lass, are not affeted by poliy-direted hanges. "U Qi U G> :: 1 «JJ 2 FC RE Fleet di tribution by in ome lass in prawlhurg by tmtegy in senario in 2. year. No vehile-holding household type is exluded from that market, irrespetive of inome group. TABLE 1 e1 Cars with Otto-Cyle :md Diesel Engines Sold :in Urban Areas During L9 by ize Market Share Total No. Car Size Engine T ype (%) Sold Small Otto ,7 3, Medium Otto ,1,927 Large Otto ,11 5,97 Van Otto 3 22,1 1 Small Diesel ,2 59 Medium Diesel 1. 13,5 Large Diesel.,771 Total 7,3,37 -u Qi U G> :: 1 1 ao 1 ona r)qj -1 High 2 OJ n-plae Group ndividual i".:l Medium EZl Low FGUR 7 Fleet distribution hy inome lass in Sprawlburg by strnlegy in senario m in 2. Sales Projetions Overview To assess the resoure and prodution impats of the vehile tehnologies defined for the TAPCUT projet it was neessary to generate estimates of new ar sales by type to urban hoa.seholds for eah of the alendar projetion years. Beause in a new ar market only the speifi utility of eah vehile type relative to its ompetitors in the same model year is of interest, a standard logit fu.ntion that inorporates the omputed utility of eah new vehile in the hoie model was used to estimate the distribution of new ar sales. The Lave-Train funtion alibrated to a 19 sales distribution (obtained from the Oak Ridge National Laboratory "Max-ket Shares" report for February 191) was used. Resulting sales shares are given in Table 1. Vehile harateristis were obtained diretly from the parameters of new modelyear vehiles haraterized for 199 and 2. Future-year demographis ( i..e., number or dis tr ibution of household ell. types) were obtained through. PF of the 1977 NPTS file distribution as augmented by national urban totals for eah demographi dimension for:east for eah TAPCUT senario in 199 and 2. Vehile totals are obtained from the PFdetermined household ell totals multiplied by the NPTS automobile ownership rate by ell, and urban vehile sales are then assumed to be equal to this total l:impfl.119 (that is, 11.9 perent of vehi les in a fleet for a alendar year are sold during that alendar year aording to t he urban vehile age distribution developed for TAPCOT). r h is proedure fores a onvention that eah househol.d type partiipates in the new ar market in proportion to the average number o f vehil es i, t held in the base Estimated sales of new diesels by size are distributed within the tehnology aording to the distribution by size of new Otto vehiles, exept as follows. Maximum sales onstraints have been imposed on diesel (and eletri) vehiles in 199. Assume that for diesels this maximum is equal to 1 perent of total sales (as the TAPCUT approah speifies for senario ) and that the utility of the diesel vehile as haraterized results in a 35 perent share in the first iteration of the model. Therefore, 19 perent has to be given bak to ompeting modes. After a finite number of iterations (usually less than eight) the diesel. share is redued to the desired 1 perent. This share is then distributed within the diesel tehnology to size lasses aording to the normalized portion of the 19 perent returned to eah Otto vehile at the end of the iterations. Thus the onstrained 1 perent takes on the harateristis of the shadow market for diesel ars represented by the unonstrained 35 perent. n man y ases this is at variane with the size-lass distribution of the Otto market, whih is often more oriented to larger vehiles (whih should be e xpeted beause the size of the benhmark diesel ar is haraterized as intermediate). Suh a result embeds the assumption that the thrust of urrent size-lass trends in the diesel vehile market will ontinue into the future. rt does not admit the possibility of a small, super-effiient diesel apturing most of the four-passenger (and smaller) market. Vehiles of nononventional or emerging tehnology that were inluded in hoie sets were not distributed by size lass after their respetive total sales shares were determined. Rather they retained the size lass of the target-market prototypes of U.S. Department of Energy researh and development programs that fous on personal vehiles. Thus a ar with a gasoline-turbine (Braytonyle) engine is always a large vehile (inluding some vans in senario under the individual travel strategy). Similarly, ars with reiproating external ombustion (Stirling-yle) engines and eletri ars are always lassed as medium-si zed (fivepassenger) vehiles. The deision not to distribute these vehiles aross size lasses was based in part on their relalively small penetration in both TAPCUT senarios and, for heat-engine ars, the late date of their market entry (making she diversifiation largely uneonomi in the TAPCOT time frame). There are also severe tehnologial problems inherent in attempting to downsi ze ars with Brayton and Stirling engines to two- and four-passenger apaities.

6 Transportation Researh Reord 952 n a few instanes the deline in holdings of ertain onventional vehile types between 199 and 2 preluded their partiipation in the 2 new vehile market. rhe sales s har ot t he se types in 2 wa s s i mply su'btrated from total ve hiles sold i no substitut e vehi l e was seleted t o fill this gap, and thus the absolute total of sales in these instanes is less than the value given by the following formulation: (.119 urban vehile totl). Senario New Car Sales The results of the personal-vehile dls _ model and s i ze-lass distr.i bution applied t o s em1rio automobiles and vans fo r eah s t r a t egy s ett i ng i n 1 9 a nd 2 a r e given i n Tables 2-. Figures - 1 show t he purhase trends over t ime by sh: l ass aross all tehnol ogies f o r eah senario strategy setting. As indiated by t he d ata iii the tabl es, unonventional-tehnology vehiles a.hi eve mode s t gai ns in s ales through 2 unde r t he i n-plae poliy and group t r avel strategy, but they perform strongly under the individual travel strategy best- tehnology fleet. Eletri ars aount for 2 and perent of ye a-2 sales, respetively, under the in-plae poliy and t he group travel strategy, whereas the -orr!! J '!'1<'1 f ng results fo r t he Brayton-yle ar are 3.7 a nd 3.3 perent. Under the individual travel strategy only the advaned batt ery-powered ar with the lithium-metal sulfide battery s ueeds among eletri a r s in ompetition with heat-engine a;::s in the year 2 ( 3.75 pe rent o f s a les). Braytons a nd Stirlings i n t he s am e yea r penet rate 9. and 15 perent of the new a r ma r ket, r e s petively. This woul d a ppear to hp r.nnslsten t with the adventurous as t of s e nario a nd the r es ul ting attitudes of both ar manufat ure r s and the ar-buying publi toward ve hi l e s i n the marketplae. As the data in Figures -1 illustrate, the sale of small ars to urban households inreases under all strategy settings, with the r a t e of inrease lowest unde the individual travel s trategy. There, medium-si zed ars lead in total sales. By ontrast, the s ales ollapse of the medium Otto-yle under the in-plae poliy, and the group travel strategy i s onsist ent with its v i rtual disappearane from TABLE 2 Projeted Urban Sales of New Cars by Size and Engine Type in Senario Under n-plae Poliy 199 Estimates" " (1 ) Car Size Engine Type Charge Charge Turbo-harge Small Medium Otto yle Otto yle. 7 J 9 l l Large Otto yle Mini Otto yle l.357 mini Small plus Diesel.22 Medium Diesel.2 Large Diesel.9 Van Diesel.35 Medium Eletri.3 (lead-aid Medium Eletri.2 (nikel-zin Large Brayton yle 3 These are manufaturing estimates, with equivalent sales assumed. btotal urban non fleet soles of new ars projeted for J 99, all sizes and engine types = O,S,. Total urban nontleet sales of new ars projeted for 2, all sizes and engine types = 11,92,. Charge Charge Turbo-harge ,.39 TABLE 3 Projeted Urban Sales of New Cars by Size and Engine Type in Senario Under Group Travel Strategy 199 Estimates',b ( ) Car Size Engine Type Charge Charge Turbo-harge Small Otto yle l Medium Otto yle l.91 l.722. OJ Large Otto yle Mini Otto yle.13 Small plus Diesel.357 mini Medium Diesel.5 Large Diesel.2 Van Diesel Medium Eletri.2.3 (lead-aid Medium Eletri.3 (nikel-zin Large Brayton yle 2 Estimates' ' ( ) Charge Charge Turbo-harge 2, l These are manufaturing estimates, with l!:uivalent sa1es assumed. btotal urban nonfleet sales of new ars projeted for 199, all sizes and engine types= 1,Sl 2,. Tota1 urban nonfleet sales of new ars projeled for 2, all sizes and engine types = 9,39,.

7 Sariks et al. 7 TABLE Projeted Urban Sales of New Cars by Size and Engine Type in Senario Under ndividual Travel Strategy 199 Estimates,b ( ) 2 Estimates,< ( ) Car Size Engine Type Charge Charge Turbo-harge Uniform Charge Stratified Charge Turbo-harge Small Otto yle Medium Otto yle Large Otto yle.5.53 Mini Otto yle.3 Small plus Diesel mini Medium Diesel Large Diesel Van Diesel Medium Eletri (lead-aid Medium Eletri (nikel-zin Medium Eletri (lithium-metal sulfide Large Bray ton yle Van Brayton yle Medium Stirling yle _ _ _793 athese ar manufaturing estimates, with equivalent sales assumed. btotal urb;.in non fleet sales of new ars projeted for 199, all sizes and engine types= 1,S,. Total urban nonoet sales of new ars projeted for 2, all sizes and engine types= J,92,. household fleet preferenes in the holdings model (Figures and 9). This is in part attributable to its unsatisfatory ompetitive position over time, speifially with referene to vehile haraterization in the stok model; that is, its market is squeezed from both sides as large ars are downweighted and small ars improve dramatially in performane. t is also partly an effet of the demographi hanges that atually our in the senario. These hanges apparently prelude reation of a viable market for a spartan general-purpose ar targeted for low-inome households. Surprisingly, medium-sized (five-passenger) ars are shown to retain the momentum of the healthy 19. Q) (/) 1,, // / //./ -----,....,,./ _ :C.-:.:: Total Sm( p) ed_(5_e) _ 9 (_:f:p)_ - Q) VJ 1 B Total Sme?.::_ d_(5_e) _ 9_(-:t:P) o r ,..., , FGURE Projeted urban sales of new ars by size in senario under in-plae poliy. FGURE 1 Projeted urban sales of new ars by size in senario under individual travel strategy. - Q) (/) :-., o -+---r.,..., r-r , r.,..., Total Sme?.::_ d_(5_e) - _( p)_ FGURE 9 Projeted urban sales of new ars by size in senario under group travel strategy. sales performane of this size lass right up to 199--with espeially strong results in the early years of the group travel strategy--in response to steeply rising fuel osts. This trend aords with enouragement of multiple-person oupany of vehiles under the group travel strategy, but it is not driven by that enouragement. However, the momentum is not sustained as household size diminishes and fast-growing operating osts ultimately move the great majority of ar owners to hoose four-passenger ars. Diesel ar sales stabilize at 5 to 15 perent of total sales by 2; they perform best, as expeted, under the individual travel strategy. The penetration is low relative to some urrent foreasts of diesel market share; this is explained by the eiling imposed on 199 sales of diesel fuel as a supply

8 Transportation Researh Reord 952.onstraint, a onstraint that effetively aps post- 199 diesel ar sales also. Diesel tehnology was represented in the vehile hoie model by a single.l!'!'te with th=' r"!nmh;nn harr:iiteristis of fiveand six-passenger diesel ars, and the diesel ars therefore ompeted primarily against Otto-yle ars in the large and medium-sized lasses. However, the progress of diesel fuel eonomy, prie, and perforff1ane is low after 199 relative to improvements in most other vehiles in the hoie model. Combined with the delining market for large ars after 199, the diesel fares rather poorly under all strategies. Senario New Car Sales The data in Tables 5-7 projet ar-size and enginetype d i stributions to vehiles in senario under eah of the three strategy settings, respetively. Figures projet sales of ars by size lass ( for 199 and 21 for the s ame t hr ee s trateqies in the same senario. Senario vehile options and level of tehnology available to buyers under the in-plae poliy and group travel strategy are idential to their senario! ounterpar ts. Tha t i s, buyers under the in-plae setting as applied in senarios and have the same new ar options to hoose from, the only differene being the re spe- tive senario new ar purhase pries, whih reflet the differing ost of finaning ( perent in senario and perent in senario ). Group travel options are also the same in the two senarios; group travel vehiles differ from those of the in-plae setting only in that the medium-sized Otto is improved in eah senario. Therefore, it is not surprising that senario foreasts (Figures 11 and ) are similar to ths for the rraspnding strategy settings applied in senario (Figures and 9). ndividual travel vehiles in senario differ from their senario ounterparts in that they represent modest aross-the-board tehnologial improvements relative to in-plae vehiles rather than signifiant new departures in tehnology or materials utilization. The medium Otto under group travel in senario is the same as the medium Otto under individual travel in that senario; all available vehiles under individual travel have improved in fuel eonomy and performane omparable to that of the medium Otto defined for the group travel strategy, but they have done so at the expense of inreased we i ght and purhase ost. Not surprisingly then, the year-2 sales distribution among the three size lasses under the individual travel strategy does not indiate the massive swing TABLE 5 Projeted Urban Sales of New Cars by Size and Engine Type in enario ill Under n-piae Poliy 199 Estimates,b (J ) Car Size Engine Type Charge Charge Turbo-harge Small Otto yle Medium Otto yle Large Otto yle Mini Otto yle.1 5 Small Diesel.31 Medium Diesel.3 Large Diesel.3 Van Diesel.3 Medium Eletri.5 (lead-aid Medium Eletri.3 (nikel-zin Large Bray ton yle 2 Estimatesa, ( 1 ) Charge Charge Turbo-harge These are manufaturing estimates, whh equivalent sales assumed. btotal urban nontlee.t S1s nr np.w r.ars projeted for 199, all sizes and engine types= 9,15,. Total urban nonfleet sales of new ars projeted for 2, all sizes and engine types= 9,3,. TABLE Projeled Urban Sales of New Cars by Size and Engine Type in Sna...'"'io Under Group Travel Strategy 2 Estimatesa, ( ) Car Size Engine Type Charge Charge Turbo-harge Small Otto yle Medium Otto yle Large Otto yle Mini Otto yle.15 Small Diesel.1 Medium Diesel.531 Large Diesel.19 Van Diesel.29 Medium Eletri.9 (lead-aid Medium Eletri.52 (nikel-zin Large Brayton yle athese are manufaturing estimates, with equivalent sales assumed. btotal urban nonfleet sales of new ars projeted for 199, all sizes and engine types= 9,15,. Total urban nonfleet sales of new ars projeted for 2, all sizes and engine types= 7,37,. Charge Charge Turbo-harge

9 Sariks et al. 9 TABLE 7 Projeted Urban Sales of New Cars by Size and Engine Type in Senario Under ndividual Travel Strategy 199 Estimates,b (1 ) Car Size Engine Type Charge Charge Turbo-harge Small Otto yle Medium Otto yle Large Otto yle Mini Otto yle.137 Small Diesel.13 Medium Diesel.57 Large Diesel.2 Van Diesel.29 Medium Eletri.37 (lead-aid Medium Eletri.2 (nikel-zin Large Brayton yle 3 These are manufaturing estimates, \Vith equivalent sales assumed. btotal urban non fleet sales of new ars projeted for 199, all sizes and engine types= 9,15,. Total urban nonfleet sales of new ars projeted for 2, all sizes and engine types= 1,275,. 2 Estimatesa, ( ) Charge Charge Turbo-harge , to small ars observed under the group travel strategy and in-plae poliy, although the emergene of a similar, if damped, trend is evident (Figure 13). whelmed by superior heat-engine tehnology in senario under the individual travel setting, in whih their penetration falls to less than 1 perent 1 1 () :i,!;; () Q) (/) Tofal Sm _e2 odj5_e) _ 9_(,,:f:p)_, Sm _e2 odj5_e) _ 9_(.:f:p)_. 2 o-r-... -r r-.-.r-r ,..., FGURE 11 Projeted urban sales of new ars by size in senario under in-plae poliy. O r-.--,..., ,-,...,...,....,...,..., ,.-, FG R 13 Projeted urban sales or new ars by size in sena.rio under individual travel strategy. () :i,!;; () Q) Cl (/) r r.,...,-r Tofal Sm _e2 <!._(5, ) _ 9_(,,:f:p)_. of year-2 sales. The six-passenger ar with a Brayton-yle engine attrats. perent of urban ar buyers under the individual travel strategy in 2; it fares less well (but redibly) under the other two strategies. Diesel, again, ahieves only 5 to 1 perent of sales aross the strategy settings in 2. Given the slightly improved tehnology set of the individual travel strategy, size-lass trends in Figure 13 exhibit a pattern similar to that under individual travel in senario (Figure 1) Large and medium-sized ars remain in the market after 2, although small ars are learly in the asendant. FGURE Projeted urban sales of new ars by size in senario under group travel strategy. COMPARSON OF TAPCUT RESULTS AND PREVOUS FORECASTS Distributions As in senario, new tehnologies ahieve modest gains in sales through 2. n senario under in-plae poliy eletri vehiles reah a 5. perent sales penetration of the urban household market. The group travel strategy, in whih total ar sales are lower, raises that penetration to 7.9 perent in 2. However, eletri ars are over- The data in Table ompare DVSAM results against ertain other published year-2 foreasts of vehile stok and sales by tehnology share and size mix. n all ases DVSAM distributions are urban-only, whereas others are nationwide. Moreover, aggreof the studies gation to size lasses in several followed proedures different from those used for

10 1 Transportation Researh Reord 952 TAPCUT. For example, Shakson and Leah (ll) used 1979 model-year body platform rather than U.S. Environmental Protetion Ageny (EPA) lassifiations to define vehile size, whih resulted in many EPA midsized ars being lassified as large by Shakson and Leah, but they were lassified as medium sized in TAPCUT. The Shakson and Leah mix-shift ase, in whih the share of large ar sales delines to 1 perent by 2 whereas midsized and small ar sales inrease, is given in the data in Table B. exlusively to four-passenger ar prodution unless this huge market is to be oneded to imports. This size mix is probably the most ontroversial result of the DVSAM and the one at greatest variane with most other published foreasts, whih envision a ontinued strong showing by the midsized and ompat (five- and six-passenger) vehiles through the end of the entury. However, none of these other foreasts were driven by fuel prie inreases as steep as those under the TAPCUT strategies. The failure TABLE Comparison of Vehile-:':iize and Engine-Tehnology Market Shares for the Year 2 in TAPCUT and Five Other Projetions (perentage of total market) Vehile Size(%) Sales Mix Fleet Mix (hoidings) Englne 1'ehnology (%! Soure of Projetjon Date Small Medium Large Small Medium Large Otto Diesel Eletri AHE" TAPCUT, senario, in-plae poliy TAPCUT, senario, individual travel strategy TAPCUT, senario, in-plae poliy Mellon nstitute ( J Lawrene Livermore Laboratory () Energy and Environmental Analysis, n. ( J) Oak Ridge National Laboratory ( J J Argonne National Laboratory ( J 5 J 1/2 1/2 1/2 / / 7/1 5/1 / _ d _ d d _ d d _ d _d d b lb 3 5 d d d 57 3 d 13 _ <l d d d d _ d d 5 J 5 d d d ;: Advaned heat-engine tehnology. bsprnwlhur only. lmports assumed by Argonne National Laboratory to be evenly divided between small and medium-sizet ars, ddata unavailable. emay inlude high-tehnology Otto share. rehnology Penetration With the exeption of the senario individual travel strategy, TAPCUT foreasts appear somewhat pessimisti on the future of light-duty diesel vehiles (in household use) relative to other published projetions. This is attributable in the ontext of year-2 sales to the superior qualities of TAPCUT Otto-yle ars ompeting with diesels: diesels retain the sales share ahieved by 199 but do not inrease that share thereafter. n ontrast, TAPCUT is relatively optimisti on eletri vehiles 1 the stok model has identified a market (generally in the low- to middle-inome range) in whih the overall harateristis of eletri ars haraterized for TAPCUT are found to be desirable. Nevertheless, operating ost and performane limitations of a ll but the very high tehnology eletri vehiles (available only in senario, individual travel strategyj inhibit signifiant market growth. Advaned heat-engine vehiles apture a high markgt share only in the senario individual travel strategy, where two suh vehiles are available in 2. n all strategy and senario ombinations, Otto-engine vehiles ontinue to aount for no less than 57 perent, and up to almost 9 perent, of sales in the year 2. Size Mix n five of the six TAPCUT strategy and senario ombinations small Otto (primarily four-passenger) vehiles dominate both fleet and sales shares by the year 2. The expliit presene of a small diesel ar in the DVSAM hoie set might have divided the small-ar share more evenly between tehnologies, but small ars would still aount for a majority of the fleet at the end of the entury. By this projetion, medium-sized and large ars ould not realistially be srapped fast enough 1 and domesti prodution lines would have to be devoted almost of the medium-sized ar in the TAPCUT projetion stems from the unsatisfatory performane harateristis of that ar relative to the ompetition in all but senario under the individual travel strategy. Final Observation Throughout the stok-modeling proess the harateristis of a vehile, rather than senario demographis, determined its fleet share and its ultimate fate. Suh marginal effets as the number of two-person households, whih appears to influene the slight differene between senario r and senario rr. under the in-plae poliy in the small ar share of the market, also play a role. But suh effets are seondary in the entral deision-making proess modeled aross the entire spetrum of households. That proess onsists of determining whether a ar is a winner or a loser rlativ'2 to the ompetition. Apparently there is no natural household market for any t ype of vehile in the DVSAM. Al - though its results may not artiulate a future plausible to all analysts, the model lear ly indiates that ar owners will ontinue to seek the best value for their personal transportation dollars, irrespetive of the soioeonomi or politial tenor of the times. ACKNOWLEDGMENT Three individuals merit speial redit for failitating the preparation of this paper. Joe Perl of Northwestern University laid the groundwork for the use in this study of the inremental logit form of the Lave-Train model in several tehnial memoranda doumenting and defending its seletion for TAPCUT. He also prepared the initial version of the FORTRAN ode for the model as used with NPTS household ell data and performed the baseline holdings distribu-

11 11 tion by ell. Mar Kaplan ontributed methodologial assistane, support, and diretion to the development of the final form of DVSAM. Sarah LaBelle assisted in early doumentation of DVSAM and provided a valuable editorial ritique of early drafts of this paper. To these persons the authors extend their gratitude. The work reported in this paper was sponsored by the Offie of Environmental Analyses, U.S. Department of Energy. REFERENCES l, S.R. Lerman and M. Ben-Akiva. Disaggregate Behavioral Model of Automobile Ownership. n Transportation Researh Reord 59, TRB, National Researh ounil, Washington, D.C., 197, pp S.R. Lerman. Loation, Housing, Automobile Ownership, and Mode to Work: A Joint Choie Model. n Transportation Researh Reord 1, TRB, National Researh Counil, Washington, D.C., 197, pp T.F. Golob and L.D. Burns. Effets of Transportation Servie on Automobile Ownership in an Urban Area. n Transportation Researh Reord 73, TRB, National Researh Counil, Washington, D.C., 197, pp L.A. Lave and K. Train. A Disaggregated Model of Auto Type Choie. Transportation Researh, Vol. 13A, 1979, pp C.F. Manski and L. Sherman. An Empirial Analysis of Household Choie Among Motor Vehiles. Transportation Researh, Vol. 1A, 19, pp T.J. Tardiff. Vehile Choie Models: Review of Previous Studies and Diretions for Further Researh. Transportation Researh, Vol. 1A, 19, pp R.H. Asin Nationwide Personal Transportation Study: User's Guide for Publi Use Tapes. FHWA, U.S. Department of Transportation, April 19.. S.J. LaBelle et al. Tehnology Assessment of Produtive Conservation in Urban Transportation, Final Report. Report ANL/ES-13. Argonne National Laboratory, Argonne, ll., Nov K. Train. The Potential Market for Non-Gasoline-Powered Automobiles. Transportation Researh, Vol. 1A, 19, pp B.R. Peterson. City Deomposition and Expansion. Report ORNL/'lM-52. Oak Ridge National Laboratory, Oak Ridge, Tenn., Sept R. Shakson and H.J. Leah. Maintaining Automotive Mobility: Using Fuel Eonomy and Syntheti Fuels to Compete with OPEC Oil. Energy Produtivity Center, Mellon nstitute, Arlington, Va., Aug L.G. O'Connelli Laurene Livermore Laboratory. Energy Storage Systems for Automotive Propulsion, Final Report--Volume : National mpat ssues. U.S. Department of Energy, De The Highway Fuel Consumption Model. Fourth Quarterly Report. Energy and Environmental Analysis, n., Arlington, Va., July D. Greene et al. Energy Savings mpats of DO;:' s Conservation and Solar Programs. Report ORNL/TM-79-v2. Oak Ridge National Laboratory, Oak Ridge, Tenn., May R. Knorr and M. Millar. Projetions of Diret Energy Consumption by Mode: Baseline. Report ANL/CNSV-. Argonne National Laboratory, Argonne, ll., Aug Projetio,n of Typial Charateristis of Automobiles and Transit Vehiles for Poliy Analysis (HARLES L. HUDSON and EV ELYN S. PUTNAM ABSTRACT n this paper the harater is tis of three future automotive tehnology sets are desribed, starting from historial data and projeted forward in time along paths suggested by given alternate future soioeonomi environments. The haraterizations inlude quantified projetions of automobile and transit vehile weight, performane, fuel eonomy, onsumer pr ie, operating ost, materials of onstrution, fuels and environmental residuals assoiated with their manufature, operating pollutants, and infrastruture-related energy expenditures, emissions, and ost. Brief desriptions of rationale.and alulational proedures are also given, and seleted results are presented. The breadth of the vehile haraterizations permits the effets of poliy options on most faets of the urban transportation setion to be examined. The methodologies developed in this work are generalized, and hene an be used with alternate assumptions in a variety of investigations. For purposes of the Tehnology Assessment of Produtive Conservation in Urban Transportation (TAPCUT) poliy analysis, eah tehnology set onsisted of six sizes of personal automobiles, eah propelled by onventional