The Impact of Socio-Economic Variables and Automobile Stock Characteristics on the Purchase of New Imported Automobiles

Rachel Dardis, University of Maryland
Diane Hrozencik, Bureau of Labor Statistics
ABSTRACT - The purpose of this paper was to investigate the impact of household characteristics and automobile stock characteristics on the purchase of new imported cars. The analysis was confined to households who purchased small cars in 1972, a year characterized by stable economic conditions and an adequate supply of small domestic automobiles. Logit analysis was used to estimate the impact of various explanatory variables. The results indicated that socio-economic characteristics, with the exception of income, had a significant impact on the probability of purchasing an imported car. In addition the origin of disposed-of-stock proved to be a major explanatory variable. Households that disposed of foreign stock were more likely to purchase imports than other households.
[ to cite ]:
Rachel Dardis and Diane Hrozencik (1985) ,"The Impact of Socio-Economic Variables and Automobile Stock Characteristics on the Purchase of New Imported Automobiles", in NA - Advances in Consumer Research Volume 12, eds. Elizabeth C. Hirschman and Moris B. Holbrook, Provo, UT : Association for Consumer Research, Pages: 538-543.

Advances in Consumer Research Volume 12, 1985      Pages 538-543

THE IMPACT OF SOCIO-ECONOMIC VARIABLES AND AUTOMOBILE STOCK CHARACTERISTICS ON THE PURCHASE OF NEW IMPORTED AUTOMOBILES

Rachel Dardis, University of Maryland

Diane Hrozencik, Bureau of Labor Statistics

[University of Maryland Agricultural Experiment Station, Article No. A-3875. Computer time was supported through the facilities of the Computer Science Center of the University.]

ABSTRACT -

The purpose of this paper was to investigate the impact of household characteristics and automobile stock characteristics on the purchase of new imported cars. The analysis was confined to households who purchased small cars in 1972, a year characterized by stable economic conditions and an adequate supply of small domestic automobiles. Logit analysis was used to estimate the impact of various explanatory variables. The results indicated that socio-economic characteristics, with the exception of income, had a significant impact on the probability of purchasing an imported car. In addition the origin of disposed-of-stock proved to be a major explanatory variable. Households that disposed of foreign stock were more likely to purchase imports than other households.

INTRODUCTION

Imports have taken an increasing share of the new car market in the U.S. since the 1970s. Their market share has increased from approximately 15 percent in 1971 to more than 25 percent in 1980 ("Domestic Car Sates"). The major reasons cited for this increase have included price, fuel efficiency and quality. However, information is limited concerning the preference for imported cars by households with different socio-economic characteristics. In addition the household's previous experience with imported automobiles has been neglected.

The purpose of this study was to investigate the factors influencing the demand for imported automobiles by households including socio-economic characteristics of households and characteristics of disposed-of-stock. The analysis was confined to small car purchasing households since most imported automobiles are small and it is important to distinguish between a household's preference for imported cars as opposed to small cars. Data were obtained from the 1972-73 Consumer Expenditure Survey conducted by the Bureau of Labor Statistics. This is the most recent data set concerning durable goods purchases and socio-economic characteristics of purchasing households.

PREVIOUS RESEARCH

There have been numerous studies concerning the demand for different size automobiles (Carlson 1976; Greenlees 1980; Lin 1982; Lave and Train 1979; Wetzel and Hoffer, 1982). Researchers have cited income, family life cycle, race, sex, occupation and education of household head, location, automobile stock and gas prices as major factors influencing the demand for automobiles.

There have been a limited number of studies concerning the demand for imported automobiles and only two studies have used disaggregate household data. The first study by Peters (1970) found that age, number of children under the age of eighteen and the number of cars owned by the household had a significant impact on the probability of purchasing imported automobiles. In addition, the number of cars owned by the household was positively related to the purchase of an imported car.

The second study by Lave and Bradley (1980) used disaggregate data collected in 1978. Their dependent variable was the ownership of imported cars rather than the purchase of imported cars. Lave and Bradley found that family characteristics (principally education), proximity to the east or west coast and number of vehicles owned by the household had a positive impact on the probability of owning an imported car.

Both studies suffered from a major limitation in that all size cars were included in the analyses. Since most imported cars are small the preference for imported cars may reflect preferences for small as opposed to large cars. Thus it is not surprising that family size and number of cars owned by the household had a significant impact on the purchase or ownership of imports.

Lin (1982) rectified this omission in his study and used state data to analyze the import share of small car sales. He found that increases in real disposable income increased import sales in contrast to the findings of Lave and Bradley (1980) and Peters (1970) who found that income was insignificant. The final study was a time series analysis of the demand for various types of automobiles including imports by Wetzel and Hoffer (1982). Sales of imports were positively related to increases in disposable income. However, the role of other household characteristics could not be investigated due to the aggregate nature of the data.

SAMPLE AND VARIABLES SELECTED

The data were taken from the 1972-73 Bureau of Labor Statistics (BLS) Consumer Expenditure Survey (Detailed Public Use Tape No. 2 and Durable Expenditures Tape). The analysis was confined to purchasers of new small cars in 1972 a year characterized by stable economic conditions (U.S. Board of Governors, 1973). The dependent variable was the probability of purchasing an imported automobile in 1972.

Selection of the independent variables was based on research reported in the literature. In addition the origin of disposed-of-stock was added as an additional explanatory variable. The inclusion of this variable was designed to reflect the impact of product experience on purchase decisions. Each of the major independent variables is discussed below.

Income

Two income measures were considered in this study. The first measure was disposable personal income, while the second measure was total household expenditures. The use of total expenditures has been defended on the basis of the permanent income hypothesis and the fact that it is a better measure of the household's permanent income than disposable income which may fluctuate over short periods of time (Prais and Houthakker 1971). Total household expenditures were used in this study since it was believed that this income measure would avoid problems associated with fluctuations in income. The effect of income was not predicted in view of conflicting research results reported in the literature.

Family Life Cycle

Family life cycle variables included age of household head, marital status, family size and number of adults in household. It was hypothesized that older households would be less likely to purchase an imported car. Six categories, ranging from less than 25 years to more than 65 years were used. Dummy variables were assigned to each category with the exception of the 35-45 age category in order to avoid producing linear dependence in the data matrix (Pindyck and Rubinfeld 1981). An increase in the number of adults was hypothesized to increase the purchase probability since this type of household might be more responsive to the price/quality characteristics of imported automobiles and more willing to experiment.

Other Household Characteristics

These characteristics included race and sex of household head, employment status of spouse and housing tenure. They were included in order to provide for variations in household tastes and preferences and life styles which might affect the demand for imported cars.

Social Class

Education and occupation of household head were used to represent the social class of the household. It was hypothesized that higher levels of education would increase the probability of purchasing an imported car. Three categories ranging from not a high school graduate to education beyond high school were used for education. Dummy variables were assigned to the lower and upper categories. Five categories were used for occupation with dummy variables assigned to each category except for the blue collar category which was omitted

Location

Four regions were reported in the survey - Western, North Eastern, North Central and Southern, necessitating the use of three dummy variables. The Southern region was the omitted region. The second location variable was the degree of urbanization i.e. whether the household lived in an urban or rural area.

Stock Characteristics

This included the number of vehicles owned and orig.n of disposed-of-stock. The latter variable had five classifications - disposed only of domestic stock, disposed only of foreign stock, disposed of both domestic and foreign stock, disposed of unknown stock, and did not dispose of stock. The omitted category was disposed only of domestic stock. It was hypothesized that households that disposed of foreign stock would be more likely to purchase imports than other households.

Purchase Price

The significance and sign of this coefficient were of interest. Thus a negative relationship between the purchase price and the probability of purchasing an import would suggest that imports were preferred to domestic models due to lower prices. In contrast a positive relationship would indicate a willingness to pay higher prices for imports due to perceived quality differences and lower operating costs for imported cars.

LOGIT ANALYSIS

Logit analysis was used to investigate the relationship between automobile size class and socioeconomic characteristics of households. This analysis is appropriate when the dependent variable is dichotomous since the predicted probabilities of purchasing a small car are bounded between 0 and 1 (Pindyck and Rubinfeld 1981). The logic probability function is given by

Pi = F (A + BXi)

    =              1           

          1 + e  - (A + BXi)

where Pi = probability of purchasing an imported car and the Xi represent the various independent variables. The individual Pi are not observed (the household purchases or does not purchase an imported car) so that it is necessary to convert the equation to an estimating form as follows.

Let                       Zi = A + BXi,

Then                     Pi = 1/ (1+e-Zi)

                         1/Pi = 1+e-Zi

                  Pi/(1-Pi) = e-Zi

and log     (Pi/(1-Pi)) = Zi

                                  = A + BXi.

The dependent variable in this regression equation is the logarithm of the odds that a particular choice (i.e. an imported car) will be made. The purchase of an imported car by a household is coded as 1 while the purchase of a domestic car is coded as 0. The method of maximum likelihood is then used to obtain estimates of A and B based on the purchase patterns of households and their socio-economic characteristics (Pindyck and Rubinfeld 1981).

The significance of all coefficients or a subset of coefficients in t he model was tested using the following test statistic:

-2 (log likelihood of the reduced model - log likelihood of the full model).

This statistic has a Chi-Square distribution with k degrees of freedom where k equals the number of variables in the full model minus the number of variables in the reduced model. Asympotetic t-tests were used to determine the significance of the individual coefficients.

The likelihood ratio index (p2) was also estimated to determine the degree of explanation provided by the model. It was calculated by the following formula:

p2 = 1 -        log likelihood of the full model    .

                log likelihood of the reduced model

The results of the logit analysis were used to estimate purchase probabilities for different socio-economic households using the equation

Pi =                    1               

              1 + e     -(A + BXi)  .

The estimated values for A and B were combined with data for various types of households to estimate the probability of purchasing an imported car by household type.

RESULTS

Thirty-nine percent of small car purchasing households bought imported cars in 1972. The mean values and distribution of the independent variables are given in Table 1. Average household expenditures were close to $12,000 while the average purchase price for a new small car was approximately $3,000.

The results of the logit analysis are given in Table 2. In the case of categorical variables the omitted category is enclosed in parenthesis. The values for the remaining categories are relative to the omitted category. A positive coefficient indicates that households in this category have a greater probability of purchasing small cars than households in the omitted category. The reverse occurs in the case of a negative coefficient.

TABLE 1

MEAN VALUES AND DISTRIBUTION OF VARIABLES USED IN LOGIT ANALYSIS

TABLE 2

ESTIMATED COEFFICIENTS FROM LOGIT ANALYSIS

The likelihood ratio statistic which is given at the bottom of the table was significant at the 5% level. This means that the entire model was significant in explaining the probability of purchasing a small car. The likelihood ratio index and the adjusted likelihood ratio index were 0.262 and 0.189 respectively. These results might be considered as satisfactory since the upper limit on the index is approximately 0.3 (Pindyck and Rubinfeld, 1981).

The asymptotic standard error is given in the third column of Table 2. The coefficient divided by its asymptotic standard error yields a t- statistic which may be used to test for the significance of the individual variables. Coefficients which were significant at the 5 percent and 10 percent level are indicated in the table.

Neither income nor purchase price were significant which is not too surprising when one considers that the analysis was confined to small cars. Age was significantly related to the probability of purchasing an imported car in two instances. Households headed by individuals aged 25 to 35 and 55 to 65 were less likely to purchase imported automobiles than households headed by individuals aged 35 - 45. The reluctance of older households to purchase imports is in agreement with research by Lave and Bradley (1980) and Peters (1970). An increase in the number of adults was significantly related to the probability of purchasing an imported car. Other significant household characteristics were sex of household heat and housing tenure.

The results for the individual social class variables were insignificant with one exception. Households headed by individuals who were not high school graduates were less likely to purchase imports than households headed by high school graduates. This result is in agreement with the findings of Lave and Bradley (1982) and Lin (1982). Location was significant in two instances. Households living in the West or in rural areas were more likely to buy imports than other households.

The most interesting result pertains to automobile stock characteristics. The origin of disposed-of-stock variables was significant in all instances. Households that disposed of foreign stock or mixed stock were more likely to purchase imports than households that disposed only of domestic stock. This result suggest that once U.S. buyers switched to foreign automobiles they were likely to repeat their purchase of such automobiles. The number of cars owned by the household was insignificant.

It is possible that the results for sets of variables such as family life cycle and social class may reflect problems of multicollinearily since the variables within each set are inter-related. Accordingly the contribution of four sets of variables was tested. These were family life cycle, social class, location and origin of disposed-of-stock. All variables sets were significant at the 5% level of significance with the exception of family life cycle which was significant at the 10% level. The results for the four reduced models are not given here due to space constraints. However, similar results were obtained to those shown in Table 2 with respect to the sign and significance of the various coefficients.

The results of the analyses were also used to estimate the purchase probabilities for different types of households and the data are given in Table 3. The representative household was a household headed by a non-black married male aged 35-45 with a high school education and employed in a blue collar job. The head was a homeowner, and lived in a rural area in the South and disposed only of domestic stock. Total household expenditures, household size, number of adults in the household, purchase price and net stock were the mean values given in Table 1.

The purchase probability for the sample as a whole and for the representative household were nearly identical. The results when changes in household characteristics are made are in agreement with the signs of the coefficients shown in Table 2. Age, sex, and education of household head, the number of adults, housing tenure and location all have a major impact on the probability of purchasing imports. In addition households that disposed of foreign stock have a purchase probability of 0.920 compared to a value of 0.367 for households that disposed of domestic stock.

The success of the logit model in predicting purchases may also be estimated by comparing actual and predicted probabilities. There was a total of 334 small car purchasers in 1972 of whom 202 purchased domestic cars and 132 purchased imported cars. According to the model the proportion predicted correctly was 75 percent. Based on the actual distribution of small car purchases the proportion predicted by chance was 52 percent (Morrison 1969). The difference between the two proportions was significant at the 5 percent level.

TABLE 3

SAMPLE PROBABILITY CALCULATIONS

DISCUSSION

The major findings of the study are as follows. First the origin of disposed-of-stock proved to be a major explanatory variable. Repeat purchases by households that owned foreign cars indicate that U.S. buyers were satisfied with the price quality mix offered by such automobiles in 1972.

Another major finding is the insignificance of the two economic variables - income and purchase price. The insignificance of income is in agreement with the findings of Peters (1970) and Lave and Bradley (1980). The latter concluded that income is only likely to be significant when the dependent variable is the number of automobiles owned by the household rather than the type of automobile owned (or purchased) by the household. The insignificance of purchase price indicates that imports were not purchased as inexpensive substitutes for domestic automobiles. Thus, factors other than purchase price must have influenced the purchase decision. According to Ward's Automotive Yearbook (1972) imports were valued for their higher quality and lack of planned obsolescence.

Family life cycle variables were significant in most instances. Households headed by older individuals, or single individuals were less likely to purchase an imported car than other households. Purchase probability increased as the number of adults in the household increased. In contrast family size was insignificant indicating the importance of controlling for automobile size class in an analysis of import demand. Other significant household characteristics were sex of household head, and housing tenure.

Both social class and location had a significant impact on purchase probabilities when tested as a whole. In addition, higher levels of education were positively associated with higher purchase probabilities which is in agreement with the findings of other researchers. Households living in rural areas were also more likely to purchase imports. This may reflect differences in driving distances, the availability of public transportation and the greater fuel efficiency of imported cars in 1972.

In conclusion, the results of this study indicate the importance of including household characteristics in an analysis of the demand for imported cars. Only small size cars should be included in such analysis. Provision for past experience with the product is also important since repeat purchases are likely to reflect satisfactory experience with the product. Data for 1972 indicate that consumers who had purchased imports were unlikely to switch to domestic small cars such as the Pinto and Vega which were introduced in 1971. This is not surprising since both these automobiles were rated as inferior to Japanese automobiles (Yates, 1983). The findings of this study are in agreement with the growing popularity of imported cars, particularly Japanese cars, in the past 12 years. According to Yates the success of imports reflects the importance of "quality, workmanship, engineering creativity, durability, as well as low operating costs" to buyers Yates (1983, p. 140). He notes that Detroit's failure in the small car market is due to emphasis on style/marketing considerations rather than on technological change and product performance. The continued superior quality of imported small cars is shown by the fact that all but one of the small cars recommended by Consumers Union in the past three years were imports (Consumers Union, 1982, 1983, 1984). At the moment the domestic automobile industry is protected by quotas on Japanese automobiles which has resulted in long waiting lines and prices higher than list prices in many dealerships (Fisher, 1984). Many of these buyers are undoubtedly repeat purchasers who, as in 1972, are unwilling to switch to inferior domestic models. The same loyalty was shown in 1982 when sates of small imported cars remained strong while sales of new cars were the lowest in 20 years (Consumers Union. 1982). [The distinction between domestic and foreign small cars is likely to diminish in the 1980s as U.S. companies respond to competitive pressure by buying small cars abroad or importing cars for sale under U.S. labels. In addition Japanese firms have established plants in the U.S. However, all these developments reflect the inability of domestic manufacturers to produce small cars that are competitive with those made abroad or under foreign management. Thus, competition between domestic and imported cars will continue though the classification of an import may change. Currently, Consumers Union provides information concerning country of manufacture and whether the firm is foreign or domestic.]

Finally it would also be of interest to compare the results of this study to results using data for the 19808. Such comparison would indicate changes in purchase probabilities over time for different households and the degree to which small foreign cars in particular Japanese cars had captured the American market.

REFERENCES

Carlson, Rodney L., (1976), "Seemingly Unrelated Regression and the Demand for Automobiles of Different Size, 1967-1975: A Disaggregate Approach," Journal of Business, 51, 243-262.

Consumers Union (1982), "CU Judges the 1982 Cars," Consumer Reports, 47, 182-193.

Consumers Union (1983), "CU Judges the 1983 Cars," Consumer Reports, 48, 180-192.

Consumers Union (1984), "CU Judges the 1984 Cars," Consumer Reports, 49, 198-212.

"Domestic Car Sales," Automotive News (1973, 1982).

Fisher, A. (June, 1984), "Can Detroit Live Without Quotas," Fortune, 109, 20-25.

Greenlees, J. S. (1980), "Gasoline Prices and Purchases of New Automobiles," Southern Economic Journal, 47, 167-178.

Lave, Charles A. and Bradley, Joan (1980), "Market Share of Imported Cars: A model of Geographic and Demographic Determinants," Transportation Research, 14A, 379-387.

Lave, Charles A. and Train, Kenneth (1979), "A Disaggregate Model of Auto Type Choice," Transportation Research, 13A, 1-9.

Lin, An-Loh (1982), "An Analysis of Small Car Sales Across States in the U.S.," Presentation at the Eighth Annual Convention of the Eastern Economics Association, Washington, D.C.

Morrison, Donald G. (1969), "On the Interpretation of Discriminant Analysis," Journal of Marketing Research, 6, 156-163.

Peters, William a. (1970), "Using MCA to Segment New-Car Markets," Journal of Marketing Research, 7, 360-363.

Pindyck, R. S. and D. L. Rubinfeld (1981). Econometric models and Economic Forecasts, 2nd edition. New York: McGraw-Hill Book Company.

Prais, S. J. and Houthakker, H. S. (1971). The Analysis of Family Budgets, (2nd ed.). Cambridge: University of Cambridge.

U.S. Board of Governors of the Federal Reserve System (1973), "1972: A Year of Accelerating Recovery," Federal Reserve Bulletin, 59, 1-11.

Ward's 1972 Automotive Yearbook (1972). Detroit, Michigan: Ward's Communications, Inc.

Wetzel, James and Hoffer, George, (1982), "Consumer Demand for Automobiles: A Disaggregated Market Approach," Journal of Consumer Research, 9, 195-199.

Yates, B. (1984), The Decline and Fall of the American Automobile Industry, New York: Vintage House.

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