63. comportamento consumo

Upload: pattinhazg

Post on 06-Apr-2018

217 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/2/2019 63. comportamento consumo

    1/8

    CONSUMER BEHAVIOR I N DIFFERENT AREASB A R R Y L E N T N E K , S T A N L E Y R. L I E BE R, A N D I RA S H E S K I N

    AB ST RA CT . Some facets of the evolution of consumer food shopping behaviorare examined by a comparison of revealed preference studies in regions of differingeconomic development: Aguascalientes, Mexico ( 1 9 6 8 ) ; Iowa ( 1 9 3 4 ) ; Iowa(1 9 60 ); and Michigan (196 6-1968). Th e majority of rural Mexicans obey theChristallerian nearest neighbor axiom, but a few patronize the capital and regionalcenters regardless of distance. This behavioral variance is a function of householdincome. Comparison of the four study areas reveals that food shopping behaviormay be universally subject to a dual assignment rule: households within a limitedrange of an opportunity exhibit a high probability of patronizing the closest place,whereas households at some distance from the nearest opportunity prefer shoppingin larger places at greater distances. The absolute range within which the Christal-lerian axiom is applicable increases considerably (from two to thirteen miles) withlevel of economic development. KEY WORDS: Aguascalientes, Central place theory,Consumer behavior, Iowa, Mexico, Michigan, Multidimensional scaling, Mult i-purpose trips, Revealed preference analysis.

    PATIAL shopping behavior is a functionS both of the characte ristics of th e decis ionmaker a nd of th e decision-making environment.These characteristics are, in turn, functions ofthe level of econom ic development. T h e rela-tionship between level of development andspatial shopping behavior is particularly pro-nounced for food shopping behavior for tworeasons. First, the primary determinants ofspatial demand (income an d mobility) and theavailability of alternative shopping opportuni-ties tend to increase with level of development.Second, the proportion of food budgets allo-cated to nonindigenous goods, which may beAccepted for publication 20 M ay 1975.Dr. Lentnek is Associate Professor of Geograp hy utthe S ta te U n iv er si ty o f N e w Y o r k i n B u f f a l o , N Y14226; D r. Lieber is Lecturer in Geography at th eUniversi ty o f Haifa in Israel; and Mr. Sheskin is agraduate student in the Department of Geography atOhio State Universi ty in Columbu s , OH 43210.

    *We gratefully acknowledge the support of theNational Science Foundation (Grant GS-2018) fordata collection and processing. Professor Gerard Rush-ton (University of Iowa) graciously made the data forthe Iowa 1960 and Michigan 1966-1968 preferencescale values available to us . Richard Buxbaum did aconsiderable amount of work both in Mexico and atthe Ohio State University above and beyond the termsof his research assistantship during various stages ofthe project in data collection and processing.

    available only at higher order centers, tends toincrease with level of development.The purpose of this article is to explore somefacets of the evolution of consumer spatialshopping behavior by a comparative analysisof research in areas differing significantly inlevel of development. The preferred investiga-tory procedure would be to obtain time seriesdata for one region, but such data are notavailable. The best alternative is to contrastcomparable studies in areas of varying develop-ment. Unfortunately, most shopping studieshave employed differing and often noncom-parable methodologies. Rushton used revealedpreference analysis to study shopping behaviorin Iowa in 1934, in Iowa in 1960, and in Michi-gan in 1966-1968.l We adopted this method-ology for an examination of food shoppingbehavior in Aguascalientes, Mexico, in 1968.Given our thesis concerning the systematiccovariation of consumer spatial behavior with!eve1 of development, we plan to:

    1 ) analyze the Aguascalientes da ta for de-terminants of revealed spatial behavior;1 Gerard Rushton, Analysis of Spatial Behavior by

    Revealed Space Preferences, Annals , Association ofAmerican Geographers , Vol. 59 (1969 ), pp. 391-400;and Gerard Rushton, Temporal Changes in SpacePreference Structures, Proceedings, Association ofAmerican Geographers, Vol. 1 (1969), pp. 129-32.

    ANNALS OF THE ASSOCIATION OF AMERICAN GEOGRAPHERS@ 1975 by the Association ot American Geographers. Vol. 65, No. 4,December 1975Printed in U.S.A.5 3 8

  • 8/2/2019 63. comportamento consumo

    2/8

    1 9 7 5 CO N S U M E R EHAVIOR 5 3 92 ) compare the spatial behavior patternsrevealed by the four studies for evidencethat spatial behavior varies systematicallywith level of development; and3 ) discuss the relationship of the evolutionin spatial behavior to the developmentof central places.

    THE A C U A S C A L I E N T E S STUDYLarge cities dominate Latin Americas urbanstructure .2 Despite the proliferation of growthpole theories purporting to explain this primatecity pattern, however, little is known empiri-cally of the function al bases of primacy in termsof consumer behavior. We hope an empiricalanalysis of food shopping behavior in the State

    of Aguascalientes, M exico, will shed som e lighton the behavioral processes leading to primacyat al l levels of Latin Americas urban structure.Data and Study A r e a

    Aguascalientes is o n the northern edge ofthe densely populated Mexican highlands (Fig.1 ) . Its urba n structure is typical of highlandMexico; i.e., the capital (Aguascalientes) con-tains almost half (160,000) of the states pop-ulation, and dom inates its trade area. T he studyarea includes all of the rural State of Aguasca-lientes except the eastern and western periph-eries, plus a small section of the Municipioof Luis Maya, Zacatecas, just to the north.Seven large towns (populations range from 55 0to 2,6 00 ) and 12 2 villages (populations rangefrom 30 to 50 0 inhabitants and consist almostexclusively of farmers and farm lab orers ) arewithin the study area.A field survey in 1 96 8 used a spatially strati-fied random sample in the seven large townsand 122 villages. A special enumeration con-ducted by the Federal Board of Elections inM ay of 19 68 provided the best d ata available,although base maps and enumeration lists con-tained considerable margins of error.3 T he sam -pling ratio of heads of households was one-half

    2 J . Miller and R . Gackenheimer, L at in A m e r i c nnUrban Policies arid the Social Sciences (Beverly Hills,California: Sage Publications, 1971), Chapter 1.

    3 The basic document used was an estimate of popu-lation of individual villages and towns published bythe State Electoral Commission, Commiss ion EstatalElectoral de l Padyon y Vigilancia Electoral (Aguasca-lientes: Delegacion Estatal del Registro Nacional deElectores, Division Sectional, 1968).

    4- I

    FIG. 1. Principal market towns of Aguascalientes.

    of one p e r ~ e n t . ~s household size averagedsix, the ratio of persons represented averagedthree percent. We believe that those interviewedrepresent a broad and fairly accurate crosssection of the population.The survey was conducted by two fieldteams, each consisting of an A merican graduatestudent and five Mexican campesinos who hadan intimate knowledge of the substantive con-tent of the questionnaires. The questionnairescontained a detailed review of the material as-pects of rural life (including items such as timebudgets and travel records). With few excep-tions, quantity and price data were fully dis-aggregated. The schedules were checked forinternal consistency and completeness, result-ing in a loss of approximately 151 of 1,01 7 in-terviews, leaving 866 consistent and completequestionnaires.5 T he variables employed in thisanalysis are :

    4 The actual sampling ratio varied between 0.35 and0.70 percent, with the bulk of the districts sampledfalling within the 0.40 to 0.60 percent range.5 A number of internal check questions were de-liberately scattered through the questionnaire.

  • 8/2/2019 63. comportamento consumo

    3/8

    BARRYLENTNEKT AL. December401) the location of each interviewees resi-

    dence;2 ) the location of predominant food pur-

    chases;3) the location of the place of work for in-terviewees receiving wages for off-farm

    labor;4) the average level of household incomefor 1967-1968;

    5 ) the average value of wage income fo r6) the average value of food purchases for

    1967-1 968;7 ) the average value of subsistence food

    consumption for 1967-1968 (evaluatedat Fall 1968 prices);

    8 ) the average value of clothing purchasedfor 1967-1968;9 ) the average value of bus fare expendi-

    ture for 1967-1968; and10) the population of all settlements.

    1967-1 968;

    Locational data are geocoded on a one-squarekilometer grid.

    Revealed Preferen ce AnalysisAs Rushton has pointed out:A fundamental description of spatial behavior is onethat describes the manner in which a populationchooses between (that is, orders) a hypothetical setof spatial opportunities; this order is place indepen-dent, that is, it is unlikely to vary significantly fromone area to another-unless major cultural bound-aries a re crossed.One method of protraying these spatial be-

    havioral patterns is by indifference surfaces re-lating to attributes constituting spatial purchas-ing opportunities. The axes of such indifferencesurfaces are commonly town size (a surrogatevariable for attractiveness) and distances ofconsumers to towns.T Each town vis-8-vis thelocation of each consumer may be measuredwith reference to these two stimuli. The con-sumer decision-making process was consideredto be a paired comparison of the nearby towns:i.e., a comparison of combinations of distancesand town sizes. By patronizing a specific town,a consumer reveals his preference for a particu-lar type of spatial opportunity relative to allother available spatial opportunities. Combin-

    6 Gerard Rushton, Preference and Choice in Dif-ferent Environments, Proceedings, Association ofAmerican Geographers, Vol. 3 (1Y71), p. 146.

    7 Rushton, op. cit., footnote 1.

    1 26004

    DISTANCE TO TOWN (KM)

    MEXICO 1968FIG. . Indifference surface for Mexican food shop-pers.

    ing the decisions of a sample of respondents, amatrix was constructed depicting the relativesimilarity of all spatial opportunities (combina-tions of attributes). Using nonmetric multidi-mensional scaling techniques, an interval scalepreference value for each combination of at-tributes was derived utilizing the rank ordersimilarities of spatial opportunities.8 The scalevalues were then used to construct an indiffer-ence surface of revealed spatial behavior. Theisolines of the indifference surfaces portray allcombinations of attributes constituting spatialpurchasing opportunities which were similar inattraction (Fig. 2 ). 9 The largest town includedin this analysis had a population of 2,600, be-

    8 Reginald Golledge and Gerard Rushton, Mul t i -Dimensional Scal ing: Review and Geographic Applica-t ions, Commissions on College Geography TechnicalPaper N o . 10 (Washington, D.C.: Association ofAmerican Geographers, 1972).

    f) The statistical basis and interpretation of revealedpreference surfaces have been discussed in R. Kernand G . Rushton, REVPREF: Paired ComparisonsAnalysis from Revealed Spatial Preference Data,Computer Institute for Social Science Research Tech-nical Report No. 95 (East Lansing, Michigan: Michi-gan State University, May 1969); J . Kruskal, Non-Metric Multi-dimensional Scaling: A NumericalMethod, Psychometr ika , Vol. 59 (1Y64), pp. 115-29;and R. N. Shepard et a/., eds., Mult i-dimensionalScal ing Theory and Applicai ions in the BehavioralSciences (New York: Seminal Press, 1972), Vols. 1and 2.

  • 8/2/2019 63. comportamento consumo

    4/8

    1975 CONSUMER EHAVIOR 541

    100-

    90-

    w 80-4g 7 0 -3 60 -u. 5 0 -

    FIG.3. Desire line map of Mexican food shopperspatronizing the capital city. The circles are three ki-lometer radii around the principal market towns.

    cause the disproportionate size of the capitalcity precluded its use in the scaling algorithm.The indifference surface reveals a dual as-signment rule for the 706 heads of householdswho were not patronizing the capital. Thelarger group patronized the nearest center re-gardless of town size, travelling less than threekilometers. This group obeyed the nearestneighbor postulate of classical central placetheory. A much smaller group (69) travelled toregional centers regardless of distance. Most ofthese consumers lived more than three kilo-meters from the nearest town (Fig. 2) . Most ofthe 160 consumers patronizing the capital re-sided at distances greater than three kilometersboth from the regional centers and from the city(Fig. 3 ) . A dual assignment rule thus may beinduced both from statistical and from carto-graphic evidence. A consumer within threekilometers of the nearest urban place obeys thenearest neighbor assignment rule; a consumerat a greater distance travels either to a largerregional center or to the primate city.Although the duality in spatial behavior be-tween short and long distances has been notedpreviously for shopping and migrational acts,the sharp break at approximately three ki-lometers warrants exp1anation.l Why do foodshoppers more than three kilometers from thenearest urban place appear to disregard dis-tance in destination choice? The answer ap-

    10 Rushton ( P r o ceed i ng s ) , op. cit., footnote 1.

    D-w MICHIGANA-A-4 IOWA

    MEXICO

    0 5 10 I5 20MILES TO TOWN7

    25

    FIG.4. Distance decay functions fo r food shoppersin three study areas.

    pears to lie in the transport modes available:one must either walk or use the extensively de-veloped bus system. The evidence suggeststhat most rural residents of Aguascalientes per-ceive the maximum walking range each way asthree kilometers. This is the outer range offoodstuffs in the classical sense of outer range.Beyond three kilometers, most prefer to utilizebus service. Once on the bus, most shoppersprefer to travel to the large centers which havea greater variety of food shopping opportunitiesat lower prices because of increased competi-tion and economies of scale.

    Economic InfluencesMost respondents (73.5 percent) purchasedfood at the same level of place in which they re-sided (Groups A and B, Table 1 . Almost

    TABLE-INCOMES OF GROUPS F MEXICANFOODHOPPERSMean family Numberincome Standard of( z scores) deviation families

    Vil lage res idents who shop inVillages (Group A ) -.125 .908 414Towns (Group D) -.222 .608 58City (Group E) -.037 .768 139

    Town residents who hop inVillages (Group C) -.618 ,101 11City (Group F) +.445 1.199 21Towns (Group B ) +.301 1.265 223

  • 8/2/2019 63. comportamento consumo

    5/8

    542 BARRY ENTN EKT A L . Decemberall of these people lived in the place of pur-chase. About one-quarter (26 .5 percen t) of thefood shoppers bypassed opportunities in theirhome settlements and patronized places atother levels (Groups C, D, E, and F,Table 1) . Th e standardized mean incomes (2scores) and sta nda rd deviations of each gro upsignificantly reduced variance in income (F =5.93 at 5 and 8 60 d.f., Ta ble 1). In general, thehigher the paired order of origin and destina-tion, the higher the level of family income, i.e.,y,. TI,

  • 8/2/2019 63. comportamento consumo

    6/8

    1975 CONSUM ER EHAVIOR 543

    2600-20-tda2 13000f0 500

    100c

    8 00020-cda3a0a

    50c

    .33

    .32

    .31.3 0

    O i i i i l bDISTANCE TO TOWN ( K M )MEX I CO 1 9 6 8

    -1.5 -1.0 75

    DISTANCE TO TOWN (MI.)IOWA 1 9 3 4

    -1.5 -1.0 75 0-* 5.0I

    0 5 10 15 20 25D I S TANCE TO TOWN (MI.)I O W A 1 9 6 0

    z0U3a0

    -ca

    P

    50c

    6000

    400020001000

    0.s#,.0

    0 5 10 15 20 25DISTANCE TO TOWN (MI.)M ICH IGAN 1 9 6 8

    FIG. . Indifference surfaces for four study areas. All non-Mexican graphs a re f rom Rush-ton, op. cit., footnotes 1 and 6 .

    1966-1968, and for the 866 households inAguascalientes in 1968. The four study areasdiffer in density of shopping opportunities,functional complexity of towns, mobility ofhouseholds (Fig. 4 ) , and level of economic de-velopment.Only limited generalizations can be based ona comparison of the preference structures ofthese four areas. The fou r studies provide arather small sample (Fig. 5 ) , and socioeco-nomic data for individual households wereavailable only in Aguascalientes. Despite these

    limitations, we believe it is useful to considerthese areas as points along a continuum of in-creasing economic development, mobility, andincome (i.e., Aguascalientes 1968, Iowa 1934,Iowa 1960, and Michigan 1966-1968).

    In Iowa Rushton concluded that the isolinesfor 1934 are almost vertical for choices madewithin eight miles of the households residence,but noted propensity to substitute larger townsize for distance at greater distances:ll

    11 Rushton ( Proc e e d in gs ) , op. cit., footnote I , p . 131.

  • 8/2/2019 63. comportamento consumo

    7/8

    544 B A R R YLENTNEKT AL. DecemberThe preference structure for 1960 shows tha t thesubsti tut ion was g reatest for choices of towns smallerthan 3,000 population and also greater for choicesmade among such towns where the ir d is tances f romhouseholds were greater than ten miles. An interest-ing behavioral implication of this is the willingnessto t ravel an extra mile , in orde r to pationize a largertown th an w ould otherwise be available, increases a sdistanc e increases.

    The inflection point was at about twelve tothirteen miles (20 km) in Michigan and atabout two miles ( 3 km) in Mexico. The studyareas may be ordered by the distance at whichconsumers convert from a stronger nearestneighbor bias (i.e., develop a strong tendencyto shop at larger centers at greater distances).Th e orde r in which th e indifference surfaces in-flect in terms of distances to towns reflects thecontinuum of development: Aguascalientes(two miles) , Iowa 1934 (eight miles) , Iowa1960 (ten miles) and Michigan (twelve to thir-teen miles). The absolute range at which mostconsumers opt for the nearest place regardlessof size increases with th e level of mobility a ndfamily income. The ability to purchase largerunit volumes on any given trip and to storefoodstuffs tends to increase with income levelsand hence a greater willingness to travel.The indifference surfaces are not directlycompa rable (except for the shape of the iso-lines) because locational types (combinationsof town size and distance) were not definedidentically. Nevertheless, a standardized fixed-effects regression model was employed to de-scribe each of these rather generalized prefer-ence patterns.12 The dependent variable wasthe scaled utility values of the different loca-tional types, and the independent variables werethe midpoints of the distance and town sizecategories for the locational types. The betacoefficients of this regression-like formulationrepresent the attribute trade-offs that con-sumers make when deciding among spatial pur-chasing opportunities. Slight changes in dis-tance values have far more overall influencetha n changes in town size. T he ratio of stan-dardized coefficients of town size divided bydistance showed a distinctly different trade-offpattern in Iowa in 1960 (0.58/0.76 = 0 .89 )and in Michigan ( 0 . 5 5 / 0 . 5 8 = 0.97) than in

    13 The awkward stat ist ical problems of using linearest imates for curvil inear functions are recognized, al-though a simple log-linear model resulted in a poorerfit ( R 2 ) than did the simple l inear model.

    TABLE-REGRESSION COEFFICIENTS FOR MODEL:PREFERENCEALUE= CONSTANT CITYSIZE+ DISTANCEStandardized

    Mexico ( 3 4 ) 0.32 1.05 (.66) -0.44 (.17)Iowa 1960 (.80) 0.57 0.58 ( . 3 5 ) -0.67 (.45)Michigan (.65) 0.37 0.55 (.31) -0.58 (.3 3)

    Area (R) Constant City size (RY) Distance (R?)

    NonstandardizedCity size DistanceArea (N) Constant (Erro r of (Error ofestimate) estimate)

    Mexico (1 8) 0.32 0.00002 -0.00232Iowa 1960 (30) 0.57 0.00017 -0.09477Michigan (3 0) 0.37 0.00016 -0.08120

    (0.00000) (0.00065)(0.00003) (0.01224)(0.00003 ) (0.01609)

    Mexico (1.05/0.44 = 2 . 3 8 ) ; rural Mexicansplaced a higher premium on distance than didrural Americans (Table 3). Mexican con-sumers had lower incomes and placed greatervalue on distance than on town size (Fig. 6 ) .C E N T R A L P L A C E DEVELOPMENT

    The trade areas for lower order centralplaces in Mexico are much smaller than in theUnited States; the range is only three kilo-meters. Household income levels are low, and

    01

    I OCMICHIGANI91

    8.

    0 (IOWA.19601

    I C AGUASCALIENTESIO0 1 ,---10 0 5 10 I 5 2 0 2 5

    RATIO OF ATTRIBUTES(Town Distance)

    FIG.6. N u m b e r of distance units per unit of townsize in relat ion to median family income.

  • 8/2/2019 63. comportamento consumo

    8/8

    1975 CONSUMEREHAVIOR 545household expenditures on food are muchlower, so Mexican central places have a muchsmaller trade volume, thus precluding retaileconomies of scale and foodstuff variety.Higher income Mexicans bypass local centersfor the primate city. The constant flow of in-come from rural areas to the regional centergenerates commercial and demographic pri-macy at the regional level, given the incomeand employment multiplier effects of retailing.Since commerce is the basic function of re-gional centers, the explanation for regionalprimacy should reflect the reasons for the con-centration of central place activities.Our limited evidence suggests that consumerchoice behavior in space is subject to lawfulevolutionary processes. In particular, foodshopping behavior may be universally subjectto a dual assignment rule. Households within alimited range of an opportunity exhibit a veryhigh probability of obeying the Christalleriannearest neighbor axiom, whereas households atsome distance prefer the more varied shoppingin relatively large centers. The absolute rangewithin which the Christallerian axiom is ap-plicable seems to increase considerably (from

    two to thirteen miles) with the level of eco-nomic development.Although we have begun to explore the re-lationship of spatial behavior to change in ur-ban hierarchical structure, we have presentedvery limited evidence, and results from otherareas are necessary. Concrete and reliable dataon threshold volumes of trade for small re-tailers in developing regions are not available,and it is difficult to specify, even stochastically,the rate of entry of intervening shopping oppor-tunities. The most reliable source for determin-ing whether an evolutionary process existswould be long term time series data for a singlearea, but such data are not available. Finally,our evidence relates only to food shopping.

    Despite these difficulties, we have attemptedto shed some light on the development of cen-tral place hierarchies. A dual assignment ruleexplains spatial behavior patterns in areas withgreat social and economic differences. Addi-tional studies using our methodological frame-work may provide the foundation for a behav-iorally based model of a hierarchical evolutionof central place.