Identification of Consumption Style Segments on the Basis of Household Budget Allocation

ABSTRACT - This paper is based on the empirical results of an interdisciplinary study that approaches consumption behavior from a holistic standpoint and analyzes a family's consumption as a part of its more general way of life. The study is based on the hypotheses that there exists systematic invariances in households' consumption and time use patterns and that households' living conditions are good predictors of these differences. These assumptions and the relevance of the new framework concept of consumption style were empirically explored with the help of household budget allocation and time use data.

Citation:

Liisa Uusitalo (1980) ,"Identification of Consumption Style Segments on the Basis of Household Budget Allocation", in NA - Advances in Consumer Research Volume 07, eds. Jerry C. Olson, Ann Abor, MI : Association for Consumer Research, Pages: 451-459.

Advances in Consumer Research Volume 7, 1980     Pages 451-459

IDENTIFICATION OF CONSUMPTION STYLE SEGMENTS ON THE BASIS OF HOUSEHOLD BUDGET ALLOCATION

Liisa Uusitalo, International Inst. for Envir. and Society, Berlin

[Ph. D., Research fellow at the Academy of Finland, Lecturer at the Helsinki School of Economics. Beginning summer 1979, research fellow at Science Center Berlin, International Institute for Environment and Society, Blissestrasse 2- D-1000 Berlin 31 (West).]

ABSTRACT -

This paper is based on the empirical results of an interdisciplinary study that approaches consumption behavior from a holistic standpoint and analyzes a family's consumption as a part of its more general way of life. The study is based on the hypotheses that there exists systematic invariances in households' consumption and time use patterns and that households' living conditions are good predictors of these differences. These assumptions and the relevance of the new framework concept of consumption style were empirically explored with the help of household budget allocation and time use data.

INTRODUCTION

Usually household consumption has been studied as an economic phenomenon separate from other activities of the family. Changes in consumption pattern have been explained mainly by economic factors, i.e., income and price changes. In addition, most empirical studies concentrate on some specific subarea of consumption at any one time and not on the total consumption structure (e.g. Prais and Houthakker 1971). Exceptions to this can be found in the analysis of consumption functions which cover the whole consumption pattern (e.g., Goldberger and Famaletsos 1970, Lybeck 1976, Carlevaro 1976) and in the time allocation theory which studies the relationship between work and consumption (Becker 1965, Muth 1966 Jacoby, Szybillo and Berning 1976). Empirical marketing oriented consumer behavior research has been dominated by the study of prepurchase decision processes and by a reductionistic approach drawing on psychology and social psychology (e.g., Arndt 1976). Consumer segmentation has predominantly been based on consumers' brand preferences or product attribute preferences.

Consumption Style Concept

This paper is dedicated to emphasize consumption as a social phenomenon, as one subfield of life where the more general features of social behavior can be observed. The paper represents au "institutional" approach which deviates from the individualistic approach in that it does not take consumers' needs and preferences as given and independent. Instead the starting point is that consumption is a part of people's way of life and that it is dependent on the existing living conditions, social and physical environment and culture (e.g. Veblen 1965 (1899), Riesman 1961 (1950), Parsons and Smelser 1956, Wiswede 1972, Nicosia and Wittkowski 1975, Mayer and Nicosia 1977, Roos and Roos 1977, Uusitalo 1979). Therefore we define consumption style as the whole of consumption activities and interests of a person or a group living under certain living conditions. (For a more detailed discussion of the concept, underlying theory and measurement problems we refer to Uusitalo 1979.)

Thus our consumption style concept differs from the lifestyle concept of most marketing studies which is often based on the psychographic profile rather than on the actual behavior of the consumer. As a result of an im-licit assumption of the independency of individual choices life style has been earlier treated mainly as an individual rather than a group property (Wells 1974, Cosmas 1976, Douglas and Urban 1977, Reynolds et al, 1977). In this paper we treat consumption style as a household property and we are in the first place interested in the realized activities of the household. Considering consumption style as a part of the way of life has many advantages. Attention can be paid to the relationship between consumption and other activities of the household, e.g., work and free time activities (Burenstam-Linder 1970, Ahrne 1976, H÷rning 1975, Gronau 1975, Ferber and Birnbaum 1977, Strober 1977, Haranne 1977, Eskola et al., 1978). Because consumption style is defined to be independent on the living conditions, it is possible to examine it against the background of the structural changes of society. As a third point we stress that the holistic approach implied by the concept allows the analysis of the functions of consumption both on a family's and on a society's level, including also the symbolic functions of consumption. [The holistic view on consumer behavior is also presented in the quality of life indicator and welfare component studies (e.g., Allardt 1975). However these studies often deal with the different resources of consumer rather than the outcome of their behavior.]

The influence of society's structural properties on consumption style can be seen in what products and services are supplied at what prices and what are the dominant consumption values in society. Structural properties have their influence also via people's living conditions, e.g., economic and intellectual resources, time, information, social position, physical milieu, etc. Within the limits of existing supply and living conditions, however, consumers' choices in the allocation of their resources have on the average increased along with the rise of income level. Consequently a household's consumption style is a result of both individual and social factors (Uusitalo 1979, 48).

Our basic assumption is that there exist systematic differences in households' consumption styles and that households' living conditions are associated with these differences. The hypothetical systematic differences in consumption activities are here called consumption style dimensions, which are to be identified. These dimensions represent the main aspects of consumption pattern according to which households differ from each other. After identifying the most important dimensions it is then possible to construct a typology of households by using these dimensions.

Earlier attempts have been made to identify empirical basic dimensions, e.g., on the basis of the households' time use (Converse 1972), leisure time activity (Karasek 1975), various life style items or household management style indicators (Arndt and Holmer 1978). Most of these studies are based on a small sample size. Previous studies do not exist on identification of consumption pattern dimensions and their explanation. However consumption has long been intuitively divided into the "necessary" and the "free choice" areas. Because the consumption standard for most people in industrialized countries has long since exceeded what is necessary for survival, the distinction is immaterial. This becomes particularly clear when we try to operationalize the distinction. Consequently a better conceptual way is needed to classify in a short way the main trends of the great variety of choices of the present-day consumer.

Operationalization of Consumption Style

Based on this very brief description of our conceptual framework we shall try with the help of empirical data to develop further the concept, that is, to explore the idea of systematic differences in consumption styles of people having different living conditions. Consumption style is here measured on the basis of household expenditures. The particular theoretical interest is in identifying invariances in consumption patterns, i.e., in the relative importance that households-under budget constraint-give to the different commodity groups. Consequently it is assumed that the allocation of expenditures rather than absolute expenditures indicate different consumption styles. If absolute expenditures would have been selected as indicators, income would have dominated as an explanatory variable because it affects the consumption of most commodity groups. However by selecting consumption shares instead of absolute expenditures, income loses its dominant position so that the relevance of other explanatory social variables can be examined. Moreover the empirical findings of time series studies show that the budget shares remain very stable in time. Thus the invariances in consumption shares would not be restricted to one cross sectional year only from which the data is gathered.

We are aware that expenditures can give only an unsatisfactory picture of the actual use of commodities or the quality and meaning of the activity connected with the commodity. Moreover consumption of free or subsidied services are included into expenditures only up to certain limit. However the nonavailability of data restricted the use of possible other indicators of consumption style (e.g., lack on time allocation data, data on attitudes).

As complementary information on activities related to consumption (as approximate measure of time use) we used about the average frequencies of leisure activities within the household. After identifying systematic differences in these free time activities separately, the key indicators of free time use were analyzed together with consumption variables.

QUESTIONS FOR EMPIRICAL ANALYSIS

We aim to answer the following research questions:

1.  Do there exist systematic invariances in consumption patterns of households, i.e., do the relative shares of different expenditures correlate in such a way that they can be described using a few main dimensions of consumption? (In other words, a factor analytical model of variation in consumption expenditure shares will be studied).

2.  How do time use variables correlate with the consumption pattern dimensions?

3.  Which of the living condition variables are significant predictors of variation in consumption style, i.e., either of those consumption variables having the highest loading on each consumption style dimension or of a household's factor score on each of the main dimensions of consumption style?

4.  What kind of consumer segments can be differentiated on the basis of the households' factors scores on the main dimensions of consumption style?

DATA AND VARIABLES

In the empirical study a household forms the research unit. The data consist of a cross sectional nationwide consumer budget survey representing the whole population of Finland. After excluding the economically inactive and one person households the final sample size was 1908. [The Statistical Central Office in Finland provided the data on consumption expenditures in its basic form (the yearly expenditure of the household on individual consumption items in Fmk). It had been collected in the year 1971 which-considering the economic cycle-happened to be a trend year (neither a year of boom nor recession).

In data collection both interviews and family book account keeping techniques have been applied. The yearly interview provided background information and information about expenditure on durables (for the whole year). Monthly bookkeeping provided information about the spending on non-durables and semi-durables. The seasonal fluctuation of these expenditures was taken into account by asking some households to act as bookkeepers every month. Households reported also the quantity of consumption of home-produce and "products of nature" as well as received natural benefits. These commodities were valued at producer price.

The basic population of the household survey comprised all private households in Finland (about one and a half million). For the monthly bookkeeping, a basic sample of households was drawn by systematic sampling from the population census. Of this basic sample all farming households were included in the final sample together with every second household in the remaining batch. The drop out percentage was 38%. The final sample size of those households participating in both the monthly bookkeeping procedure and the yearly interview was 3512. In our study only economically active families were selected as an object of study. Therefore the sample of the household survey was further restricted by excluding from socioeconomic groups: the "economically inactive" (retired, students) and from household types: the one-person households, the one-parent households with children and the households with four or more adults.

The sample was compared with the figures of total population register. The farming households were somewhat over-represented in the sample. (For a more detailed discussion about the validity and reliability of data see Uusitalo 1979, 64-69).]

For the purpose of identifying consumption pattern dimensions, 220 consumption items or consumption group items were selected as variables from the list of household budget survey. (Certain rules were applied in the selection in order to avoid technical correlation between variables.) This set of variables was then further restricted on the basis of the intercorrelations between the variables. For technical reasons the analysis was started by doing it first for three subgroups of variables (I Food, II Clothing, transport, household, health, III Leisure consumption), and then by selecting the best variables for the analysis of total consumption.

Time use variables (20) were constructed from the leisure activity frequencies of individual household members (children under 10 years were not interviewed). For each household an average figure was calculated to express the activity level of the household.

The original set of 30 explanatory living condition variables was restricted by principal component analysis and by selecting only one variable on each component (the components were: income, family size, type of living milieu and housing, standard of housing, socioeconomic status, work of spouse, life cycle stage). This was made in order to avoid intercorrelations between the explanatory variables. Dummy variables were constructed for regression analysis. In multiple classification analysis the original classificatory variables could be used after certain reclassification operations.

ANALYSIS AND RESULTS

Identification of the Main Structural Dimensions of Consumption.

Consumer Style dimensions were identified first for three subgroups of variables and then for a final group of variables representing the total consumption. The number of dimensions, i.e., the number of factors in the factor analysis was defined on the basis of the eigenvalue and of the proportion reduced by the factor.

In the final analysis three main dimensions emerged which met the criteria and which were easy to interpret. [The results of the subanalysis as well of a larger set of variables representing the total consumption were also helpful in the interpretation of the dimensions.] These three factors accounted for 32% of the invariance of the final set of 20 selected variables representing various subareas of consumption (Table 1). The three main dimensions were interpreted as MODERNITY, VARIOSITY and MOBILITY (Table 2). All three dimensions had two poles but were labeled according to one pole in order to simplify the presentation. The results were validated in different subsamples of the data (based on the living milieu, income, age and socio-economic strata) which all gave identical main dimensions.

TABLE 1

VARIMAX ROTATED FACTOR MATRIX

TABLE 2

ALL EXPENDITURES (20 VARIABLES) FACTOR INTERPRETATION BASED ON THE HIGHEST FACTOR LOADINGS

MODERNITY (modern vs. traditional consumption) appears as the first common factor that accounts for most of the variance. MODERNITY of consumption mainly reflects the use of modern "far-processed" products and services (e.g., manufactured food, restaurant services, alcoholic beverages) as opposed to the more traditional form of consumption with a high degree of home production. Other items on this factor refer also to a modern urbanized way of life (Public transport, rented dwelling). However it must be pointed out that MODERNITY in consumption does not require city surroundings; in smaller centers or even in the country a modernized style of consumption can also become prevalent. In the same way, a traditional, home production associated consumption may become popular in city-surroundings within certain groups.

The second factor MOBILITY indicates the relative amount spent on a private car. The opposite pole contains either necessities or dwelling expenditure. Those having high share for food expenditure, i.e., low income households have low MOBILITY. On the other hand those households that spend a large proportion of their budget on own house or dwelling cannot afford a high share for car expenditure at the same time (or they have in earlier years invested in a car).

The third factor dimension VARIOSITY indicates the relative amount spent on enrichment of consumption basket (recreation, education and culture, special food, fruit, vegetables, clothing, furniture etc.). The opposite pole consists of necessities. [For example, if the share of newspaper subscription of the total leisure consumption is very high this means that no other leisure consumption exists, i.e., low VARIOSITY.] This dimension is closest to the old distinction between necessary and free-choice consumption although that distinction is also to a certain degree built into the two other main dimensions.

The above mentioned three dimensions were the main underlying structural dimensions in consumption pattern. But it is worth noticing that a considerable part of variation in consumption patterns still remains unaccounted for by these common dimensions. The relatively low percentages of the total variance explained could be attributed to a number of causes. We refer first to the measurement errors of the large survey data. Secondly we infer that the budget shares for many consumption items show random variation. This seems to be true especially of several specific leisure items and, e.g., household durables. It is also conceivable that items on which the share of spending does not show systematic differences are evenly accessible and accepted (or very rare) in all population groups. It is also difficult to establish the differences in budget shares for individual items the yearly budget share of which remains very small for all households. It is worth adding that consumption patterns can vary according to numerous different premises. It is therefore possible to predict that a few dimensions are able to account for only a part of the total complexity.

Irrespective of these precautions, we may state that the identification of the MODERNITY, VARIOSITY, and MOBILITY dimensions of consumption patterns affects to a noticeable degree the possibility of conceptualizing (giving empirical content) and comparing the different consumption styles. By using the concept of consumption style we now refer to the three main aspects of consumption: (1) how modern (mass product and services directed) vs. traditional (home production directed) it is, (2) how many-sided and various vs. one sided and "poor" is the content of the commodity basket and (3) how big a role private transport plays in consumption pattern. The analysis also helped to find the most relevant separate indicators (consumption items) of consumption style.

Correlation of Time Use Variables with Consumption Pattern Dimensions

The time use patterns of the household were examined similarly by doing factor analysis on time use variables. The best indicators of systematic variation in time use were then analyzed together with consumption variables. The best indicators of systematic variation in time use were then analyzed together with consumption variables in order to see how they correlate with each other. Not surprisingly, modernized consumption was found to be correlated with such free time activities as bar/disco/ restaurant visits and cinema visits, and traditional consumption with total household work at home (of wife) and church attendance. All cultural activities (e.g., book reading, theater, concerts) as well as sport and association activity correlated with variosity in consumption pattern indicating an enriched way of life also in other aspects than consumption. No one of the time use variables correlated with the mobility-dimension.

The Influence of Living Conditions

The next step was to find those living condition variables that have most influence on how high a place the household will have on the dimensions of MODERNITY, VARIOSITY and MOBILITY respectively. This problem was approached by means of three different methods which all gave very consistent results as far as the best predictors and the direction of their influence were concerned.

First-in order to have a general picture of which living conditions are related to each dimension of consumption-a combined factor analysis was made where the living condition variables and the selected time use variables were added in addition to the consumption variables. This way of using factor analysis in causal interpretation is not very common but it is consistent with the idea of the way of life approach where the "whole" of activities and the conditions under which they take place should be considered simultaneously rather than by making a clear distinction between dependent and independent variables. The most important living conditions and time use variables for each consumption dimension were then identified on the basis of the factor dimensions on which they received the highest loadings.

The second method of exploring the influence of background variables were to make several stepwise regression analyses where the dependent variables were those consumption variables that had a high loading on each dimension: in other words we used the best indicators of each consumption dimension to represent the whole dimension. A linear additive model was assumed of the dependency of the consumption share on the explanatory living condition variables.

The third method was to explain directly the households' factor scores on each dimension. This was done by both multiple regression and MCA-analysis. The latter gives information on both the relative importance of the explanatory variables and also gives more precise information on the direction of dependence in different subclasses of the explanatory variables. In MCA there is no need to assume a linear relationship between the dependent and explanatory variables. Table 3 gives the results of MCA, i.e., the factor scores' deviation from the grand mean in the different classes of the predictor variables. Table 4 summarizes the results of the different analyses of dependence.

We can say by way of summary that the most important predictors of MODERNITY are: not being a farmer, living in a city or other population center and being young. A working spouse and high standard of housing (but small living surface) are also related to high modernity in consumption.

This dimension has become increasingly relevant along with industrialization and the migration of population from farming communities to population centers. Changes in the age structure of the population (e.g., the post war population explosion reaching adulthood in the 70's) may also be reflected in the degree of MODERNITY in consumption (e.g., in the relevance of restaurant services, in the use of manufactured food instead of home-produced food, and in the use of alcoholic beverages and urban forms of entertainment).

The best predictors of VARIOSITY are: having a white collar occupation (and high education level) and high income. In addition, a high standard of housing, a high existing stock of durables, the city as a living milieu and youth correlate with the above mentioned predictors.

The occupational and income structure as well as the education level are structural properties of the society the changes of which are reflected in the VARIOSITY (enrichment) of consumption. The increased VARIOSITY means that, e.g., the spending on recreation, education and culture, enrichment of food, household furniture and fittings will increase. Time use for cultural activities and associations also forms an element of VARIOSITY.

A considerable part [As Allardt (1976) has pointed out, in social research it is common that statistical models-assuming that the variables are not trivial nor logically dependent on each other-explain only a small part of the variance of the dependent variable. Often 30% is considered a "good" percentage. To a certain degree this is due to the characteristics of social life where no mechanistic causality can be found and where people often make different choices under similar conditions.] of the variation of MODERNITY and VARIOSITY can be explained by the living condition variables. The multiple R-squares were in MCA-analysis 26 and 37 respectively. (In the separate regressions of the best individual consumption style variables multiple R-squares varied between 17 and 29%.)

TABLE 3

SUMMARY TABLE OF MODERNITY, VARIOSITY AND MOBILITY-FACTOR SCORES' DEVIATION FROM GRAND MEAN IN THE DIFFERENT CLASSES OF THE PREDICTOR VARIABLES

In the case of MOBILITY the result is different. MOBILITY (the relative amount spent on a private car) is rather independent of any other living conditions than income. The R-square was in MCA 5.6% and in the regression analysis even lower. This factor dimension may well be peculiar to some European countries where car taxation and gasoline prices are deliberately kept at a relatively high level. This makes the budget allocation differences very considerable between those households having a high car expenditure and those not having a car as far as consumers at the lower or medium income level are concerned. On the other hand, the decision whether to have a car or not seems to be independent of any other living condition variables than income. This refers to the symbolic importance given to a car in all population groups. Even income is not a very good predictor of car use or car ownership. Income explains about 3% of allocation on transport and 9% of car ownership. Other living condition variables did not prove to be very relevant although age (especially under 30) seemed to increase somewhat the eagerness for spending on a car.

TABLE 4

SUMMARY OF THE BEST PREDICTORS OF MODERNITY, VARIOSITY AND MOBILITY IN THE DIFFERENT ANALYSES

Because socioeconomic status proved to be the most relevant predictor for both MODERNITY and VARIOSITY in consumption, a comparison was also made of the location of different socioeconomic groups on the dimension map. The results supported the previous interpretation of the main dimensions of consumption. On the MODERNITY-dimension Farmers were differentiated from other, more urbanized groups whereas distances between the urban groups were small.

There were very considerable distances between all socioeconomic groups on the VARIOSITY-dimension: the group Managers and higher administrative was located highest, next came Clerical and Service workers. Farmers and Unskilled workers were low on VARIOSITY. On the MOBILITY-dimension there were no considerable differences between the socioeconomic groups. [For a more detailed analysis of socioeconomic group differences in consumption style see Uusitalo (1979), 121-126 and 196. For comparison of social stratification and consumption pattern see Munters (1977) and Uusitalo (1977).]

Constructing a Consumer Typology with the Help of the Dimension Scores

On the basis of the household's factor scores on the three dimensions it was classified in one of the eight possible consumption style segments (I-VIII). (A factor score below the total sample mean was classified 'low' and above it 'high'.) Enough households fell in each segment so that all of them were subjected to further analysis.

FIGURE A

DISTRIBUTION OF THE SAMPLE IN DIFFERENT CONSUMPTION STYLE SEGMENTS

Figure A shows the household distribution among the segments and the mean total available income and total expenditure of each segment. The average amount of saving has also been calculated based on the income and expenditure figures. In order to find out more of the characteristics of each consumer segment, we examined first how they differ in their consumption and also how they differ in their living conditions.

Figure B shows the budget allocation differences between the segments. The budget allocation of each segment differs considerably from the population average. When analyzing changes in consumption structure it seems to be important to consider also into which segments households have recently moved or are moving instead of merely stating average changes in expenditure shares.

FIGURE B

THE ALLOCATION OF CONSUMPTION EXPENDITURE IN DIFFERENT CONSUMPTION STYLE SEGMENTS

The four first segments are high on MODERNITY, which is reflected in high allocation for "Other goods and services''. Segments I-II and V-VI are high on VARIOSITY, which can be seen from their relatively high allocation for "Recreation, education and culture" and "Furniture and fittings".

Segments I, III, V and VII are high on MOBILITY and the figure shows how their allocation for "Transport" is on a much higher level than in the other segments. In high-spender (high income) segments the low amount of total expenditure spent on food allows the high share for "Transport''. In the medium income segments III and VII, in turn, it requires a lower share also, e.g., on "Recreation, education and culture, "Furniture", "Clothing" and "Dwelling''. This fact becomes even more clear when we compare, e.g., the segments II and III which have about the same income level and degree of MODERNITY but different MOBILITY (=different allocation for Transport). The comparison of the traditional segments VI and VII gives similar results.

Table 5 gives the index of average absolute expenditure in each segment after relative shares have been converted back into absolute figures. This was made in order to get a more concrete picture of the absolute differences in consumption level between the segments.

TABLE 5

INDEX OF AVERAGE ABSOLUTE EXPENDITURE IN DIFFERENT CONSUMPTION STYLE SEGMENTS

Based on a detailed analysis of the consumption, time use and background characteristics of each segment a tentative typology was proposed by giving a name to the different segments (Figure C). The four main consumer types are defined on the basis of dimensions MODERNITY and VARIOSITY, which are more general dimensions describing consumption style differences. Each of the four basic types are then further divided into two types on the basis of the MOBILITY-dimension. Some of the main features of each segment which were found by detailed comparison of segment profiles are repeated beneath the figure.

The arrows in the figure are very speculative too. They present the writer's intuitive view of the most probable changes from one segment to another. It has often been claimed that there is an increased tendency of absence of roots (not having meaningful goals and content in life) of those moving from traditional, rural consumption culture to an urbanized way of living. Some of the suppressed urbanized consumers may also easily follow the "alienated" consumption model which means that the increased income will be spent on car, restaurant and alcohol. It can be noticed that among the segment "alienated", e.g., the expenditure for own dwelling (apartment or house) is lower than for any other group although their standard of housing (small houses in outlying districts of city or scattered houses) is among the lowest. Of course the change from the traditional rural and marginal models may also happen in the direction of either the enriched traditional model or in the direction of the active urban model. It is probably psychologically least easy to move from the "alienated" segment to any other consumption style segment.

FIGURE C

CONSUMER TYPOLOGY

The figure also shows another tendency, the tendency of "neotraditional" behavior which in Europe has gained popularity in the last years: to favor a more traditional and simple, close-to-nature way of life (moving into the country or small suburban houses, small scale farming or gardening, home production of food and clothing, increased sport activity and activity connected with the nature). However this differs from the traditional rural model in the higher standard of housing and equipment and more varied free time activities. At the same time several farmers have reached the same standard of modern rural life and thus belong also to this "active home production favorizer"-group.

The classification of households into differing consumer segments was made for practical purposes too. The emerged segments, "consumer types", give possibilities both for diversifying marketing measures for these target segments and for identifying especially "problematic" consumer segments for social policy or consumer policy planning.

Segments I and II are ideal consumers: from the point of view of the marketing of manufactured consumer products as well as cultural products and services. Recreation and home decoration markets find their potential customers particularly in segments I, II, V and VI. Car and durable marketers can (in addition to the previous segments) count on segments III and VII as well. Restaurant and alcohol customers are found in all urbanized consumer types (I-IV) although the social problems of misuse accumulate mainly in groups III and IV.

Consumer and social policy problems again originate on the basis of the situation of the suppressed segments, e.g., in the case of the urbanized poor the problem is that they do not only spend relatively the largest share on food but also do this absolutely. This can depend on either lower economic ability [See (e.g., Michael 1972, +lander 1978, Joerges 1979)] or the fact that spending on food is the first type of "luxury" that families can afford. Also health care problems arise due to the lack of nutrition in the one-sided food consumption of the urbanized or rural segments III, IV, VII and VIII, or of the excessive fat, sugar and milk product consumption of farmer households.

Also more far-reaching questions concerning social welfare may be raised on the basis of the findings. What would be a recommendable balance between "modernized" and traditional forms of household consumption; e.g., a balance between manufactured product-centered consumption that involves productive activities of the household itself? Should a socially active way of life be encouraged rather than a family-centered "privatized" way of life? Can total consumption or the variosity in consumption and time use be more equally distributed among households? What should be done in favor of those consumption is linked with a passive, inactive use of time?

CONCLUSIONS

The major contribution of the paper lies in identifying the main structural dimensions of consumption style, thus giving empirical content to the concept and new criteria according to which the differentiation of households' consumption and time use patterns can be described and analyzed. The findings support theoretical assumptions according to which consumption behavior of a house-hold-when taken as a whole-is also dependent on other social determinants than income. In applied consumer theory the social determinants of behavior are often excluded from the analysis, because for most only certain limited consumption choices are examined (product or brand choices). Social determinants have not proved to be good predictors in explaining this limited, operative choice behavior of the household. However the more we approach the "strategic" choices of the household (extensive decision making), the more important the role of social determinants becomes in these choices.

Socioeconomic status, age and living milieu proved to be most relevant in explaining consumption style differences measured by budget allocation. Of course income is correlated with all the above mentioned factors and is thus included in the relevant factors. However, income alone was a satisfactory predictor of only one main aspect of consumption style, i.e., MOBILITY (expenditure for private transport). In previous studies several differences, e.g., in the enrichment of consumption have been summarized under income influences. Now they are rather specified as result of way of life differences between so called white collar status and other occupations. Thus the clearest combined effects of social status and income can be seen in the VARIOSITY in consumption style. The clearest combined effects of social status, living surroundings and age is reflected in the degree of MODERNITY in consumption style.

The composition of a household and its life cycle stage has earlier gained considerable attention in explaining consumption differences. However these life cycle and family size factors did not prove very relevant in our study. [The influence of age (on MODERNITY) seems to refer to consumption style differences between generations rather than between different life cycle stages.] There seem to be basic differences in households' behavioral patterns which are related more to other social determinants than these factors.

Due to the fact that the study consisted of cross sectional data, no exact relation could be established the changes in living conditions and consumption style. However when looking at the different consumption style segments it becomes evident that an increase in income can affect consumption style very differently depending on whether it is combined with a rise in education level or not, or if it happens for a household which is already an established city (or country) dweller or for a household moving, e.g., from the country in the city.

REFERENCES

Ahrne, G÷ran (1976) Den gyllene kedjan - studier i arbete och konsumtion. Falk÷ping. Prisma

Allardt, Erik (1976) Hyvinvoinnin ulottuvuuksia (Dimensions of welfare). Porvoo/Helsinki. Werner Soderstrom Osakeyhti÷ (also in Swedish: Att ha, att Slska, att vara. Lund 1975).

Andrews, F. M., J. N. Morgan, and J. A. Sonquist (1967), Multiple classification analysis, Survey Research Center, I. S. R. University of Michigan, Ann Arbor

Arndt, Johan (1976), "Reflections on Research in Consumer Behavior," in Anderson, B. (ed.) Advances in Consumer Research, Vol. III, Cincinnati, Association for Consumer Research, 213-21

Arndt, Johan and Holmer (1978), "Dimensions of Household Economic Management," Journal of Consumer Studies and Home Economics, 2, 27-34

Backer, G. S. (1965), "A Theory of the Allocation of Time," Economic Journal, (September 1965)

Burenstam-Linder, S. (1970), "The Harried Leisure Class," New York, Columbia University Press

Carlevaro, Fabrizio (1976), "A Generalization of the Linear Expenditure System," in Solari, L and Du Pasquier, J. N. (ed.), Private and enlarged consumption (ASEPELT 5), Amsterdam, North Holland Publishing Co.

Converse, Philip (1972), "Country Differences in Time Use," in Szalai (ed.), The Use of Time, 145-177

Cosmas, S. C. (1976), "The Advantages and Disadvantages of the Profile Approach to Analyzing Life Style Data," Advances in Consumer Research, Vol. III, Association for Consumer Research, 501-503.

Douglas, S. P. and C. D. Urban (1977), "Life Style Analysis to Profile Women in International Markets," Journal of Marketing, 41, 46-65

Eskola, K., M. Kortteinen, and J-P Roos (1978), "The Way of Life and Cultural Activities in the Finnish Family," University of Joensee. Research Reports 4

Ferber, M. A. and B. G. Birnbaum (1977), "The 'New home economics': Retrospects and Prospects," Journal of Consumer Research, Vol. 4, (June 1977), 19-28

Goldberger, A. S. and T. Gamaletsos (1970), "A Cross-Country Comparison of Consumer Expenditure Patterns," European Economic Review, Vol. 2, Spring 1970, 357-380

Gronau, R. (1973), "The Intrafamily Allocation of Time: The Value of Housewives' Time," American Economic Review, Vol. 63, 634-51

Haranne, M. (1977), "Leisure Time, Work and Cultural Activities in Present Finnish Daily Routines," Research reports 212, Department of Sociology, University of Helsinki

Homing, K. H. (1975), "Remarks on Interest Orientations in Work and Non-work Situations," Berlin. Wissenschaftszentrum. First workshop on consumer action research

Jacoby, Jacob, G. J. Szybillo, and C. K. Berning (1976), "Time and Consumer Behavior: An Interdisciplinary Overview," Journal of Consumer Research, Vol. 2, (March 1976), 320-339

Joerges, B. (1979), "Die armen zahlen mehr - auch fnr Energie," Zeitschrift fnr Verbraucherpolitik 3, No. 2

Karasek, A. (1975), Patterns of Leisure, unpublished paper, Institutet f÷r Social Forskning, Stockholm

Lybeck, J. A. (1976), "The Allocation of Consumer Expenditures in a Swedish Macroeconomic Model," European Economic Review, Vol. 8, (December 1976), 371-

Mayer, R. N. and F. M. Nicosia (1977), "The Sociology of Consumption: Searching for Useful Perspectives on Society,'' European Research, 5, (July 1977), 143-152

Michael, R. T. (1972), The Effect of Education on Efficiency in Consumption, New York, Columbia University Press

Munters, Q. J. (1977), "Social Stratification and Consumer Behavior," The Netherlands' Journal of Sociology, 13 (1977), 153-173

Muth, R. F. (1966), "Household Production and Consumer Demand Function," Econometrica, Vol. 34, 1966

Nicosia, F. and T. H. Wittkowski (1975), "The Need for a 'Sociology of Consumption' in Sternthal, B. and Zaltman, G., "Broadening the Concept of Consumer Behavior"

Parsons, Talcott and Neil J. Smelser (1956 and 1966), "Economy and Society: A Study in the Integration of Economic and Social Theory," London, Routledge and Kegan.

Prais, S. J. and H. S. Houthakker (1971), "The Analysis of Family Budgets," The University Press, Cambridge.

Reynolds, F. D., R. C. Melvin, and W. D. Wells (1977), "The Modern Feminine Life Style," Journal of Marketing, 41, 38-45

Riesman, David (1950 and 1961), The Lonely Crowd: A Study of the Changing American Character, New Haven/ London, Yale University Press

Roos, J-P and B. Roos (1977),VSlfSrdsforskning och levnadssSttforskning, Sociologisk Forskning XIV (1977), 1, 1-13, (Welfare research and way of life research)

Sternthal, B. and G. Zaltman (1975),"Broadening the Concept of Consumer Behavior," The Association for Consumer Research, 1-24

Strober, M. H. (1977), 'Wives' Labor-force Behavior and Family Consumption Patterns," American Economic Review, Vol. 67, No. 1, 410-417

Uusitalo, Liisa (1977), Konsumtionsstilen som en del ay levnadssSttet, Publications of Helsinki School of Economics, D:22

Uusitalo, Liisa (1978), Konsumentinformation some f÷rutsSttning for och formare ay behovstillfredstSllelsen, in Konsumenten i samhSllet, 90-97, Stockholm, Raben and Sj÷gren

Uusitalo, Liisa (1979), "Consumption Style and Way of Life: An Empirical Identification and Explanation of Consumption Style Dimensions", Publications of the Helsinki School of Economics A:27

Veblen, Thorstein (1899 and 1965), "The Theory of the Leisure Class," The MacMillan Company, London

Wells, W. D. (ed.) (1974) Life Style and Psychographics, American Marketing Association, Chicago

Wiswede, G. (1972), Soziologie des Verbraueherverhaltens, Stuttgart, Ferdinand Enke Verlag

+lander, F. (1978), "The Effects of Income Level upon the Economy of Buying, or do the Poor Pay More," Paper presented at 3rd European Colloguim on Economic Psychology, Augsburg, July 27-29, 1978

----------------------------------------