Consumption theories and consumers' assessments of subjective well-being.
The following quotes provide an introduction to the main ideas for three different models of the relationship of changes in consumer resources to perceptions of well-being.
Each of these statements posits how specific changes in resources and/or preferences affect well-being. Yet all have been ignored in the recent expansion of research on the consumer life cycle. Attempts to understand variation in family and consumer well-being over the life cycle have emphasized the specification of variables to capture the relevant demographic characteristics of each stage in the life cycle (Murphy and Staples 1979; Stampfl 1978). A growing literature on subjective assessments of well-being has been motivated by concern for the sources of reduced satisfaction at particular stages (Campbell, Converse, and Rodgers 1976). Analyses of the implications of changing economic and demographic conditions for economic well-being also have emphasized life cycle stage as a critical variable (Levy 1987).
Although variation in consumption needs and resources to meet them has been considered extensively, no systematic empirical effort has been made to relate subjective well-being assessments to economic theories about life cycle consumption behavior. This conclusion is supported by the fact that the literature on life satisfaction has relied on a single, short term measure of economic resources, current income, in conjunction with demographic variables to capture other influences of stage in the life course (e.g., Allardt 1977; Andrews and McKennell 1980; Dohrenwend 1973; Fernandez and Kulik 1981; Gray et al. 1983; Hall 1976; Haring, Stack, and Okun 1984; Keith 1985; Kessler 1982; Link 1982; Mutran and Reltzes 1984).
The research presented here is part of a larger project devoted to improving the specification and investigation of relationships among a more comprehensive array of objective economic variables and subjective well-being indicators than heretofore. Prior work on subjective well-being has taken as implicit that current income translates directly into resources for consumption. There has been very little recognition of theories acknowledging the divergence between consumption and income that results from the need to save, the ability to borrow from future income, or pressures to dissave to meet perceived or actual increases in consumption needs. However economic theories and empirical studies of savings and the consumption function have repeatedly demonstrated that utility maximizing consumers do not consume all of their current income (Modigliani and Brumberg 1954) and that current income may be a poor indicator of current consumption as well (Friedman 1957). Other components of financial security undoubtedly contribute to a sense of well-being. For example, individuals with annual incomes of $30,000 are highly likely to diverge with respect to other financial resources. Some have low or negative net worth (i.e., assets minus liabilities), while others have relatively high net worth. Some have relatively inadequate contingency assets. On the other hand, many may be experiencing relatively high, while others relatively low, current incomes with respect to future incomes.
While a positive relationship between current income and psychological well-being has been documented extensively, there is also evidence of important exceptions. Campbell (1976) reviewed five national surveys, which all revealed that the proportion of "very happy" people increases with income level. However, a large minority of the affluent describe themselves as less than very happy and a substantial minority of the least affluent claim they are very happy. Thus accurately evaluating the economic resource relationship to psychological well-being would seem to require specification of additional objective economic measures, guided by appropriate economic theories.
The next section relies on the insights of Modigliani, Duesenberry, and Kyrk to develop objective economic measures for life cycle income, relative income, and of deficits in resources to meet individual consumption aspirations, respectively. After explaining the nature of the data set, separate regression models are specified consistent with each of the theoretical economic well-being hypotheses. Then the estimated regression results are used to assess the relative explanatory power of each measure on respondents' subjective sense of overall well-being and to obtain conclusions relevant for current research on life cycle effects.
THEORIES OF ECONOMIC WELL-BEING
According to Duesenberry's theory of consumer behavior, "interdependent preferences" mean that the consumer's satisfaction depends on the level of own consumption, as perceived relative to others'. Specifically, "when an individual makes an unfavorable comparison of his living standard . . . the individual is dissatisfied with his position" (1949, 32). Furthermore, the interdependence is connected to the motivation for saving because the dissatisfaction "arises from the rejection of impulses to spend" (1949, 32). It follows that those who are able to consume at or above the same level as their peers should be more satisfied than those who cannot, provided that they are also able to save enough to be satisfied with that position. This important caveat regarding savings explains how relative consumption comparisons are linked both to current and future income levels. Additionally, it makes clear that the relative income hypothesis (RIH) and Modigliani's life cycle theory of saving (LIH) are closely related alternatives for explaining how satisfaction depends on both current and expected incomes. Modigliani (1986) acknowledged the relationship of his work to Duesenberry's by referring to the "Duesenberry-Modigliani" consumption function in his Nobel address.
One interpretation of the RIH that relates present to future consumption is as follows. Those consumers who are unable to "keep up with the Jones's" experience dissatisfaction as pressures mount to increase current consumption above the level which is consistent with their desired level of future consumption. Empirically, this hypothesis can be examined by relating current household expenditures to those of other households during the same period, holding income constant. Thus, relying on current data alone, one can test a hypothesis about the relationship of current vis-a-vis future resources as determinants of current satisfaction. Although this type of test requires complete expenditure data, it avoids the notoriously difficult problem of obtaining reliable data on savings.
Modigliani's LIH framework concerns optimal savings behavior that depends on and determines differences between current income and expected future income. In its simplest "stripped down" form, the LIH posits life cycle paths for consumption, saving, and wealth that depend on the length of the work life, retirement age, and expected age at death, as well as the age profile of earnings (Modigliani 1986). Because utility maximization requires consumption smoothing over time, the path of wealth holding is hump-shaped. During the earning phase of the life cycle, households save to accumulate wealth that is later "spent down" in retirement. Thus the LIH formulation holds that current consumption depends on the entire income stream across all ages. Even without transitory income variation,(1) current income may differ substantially from the level expected in subsequent periods because of one's particular age-earnings profile. Thus to examine Modigliani's specification, the estimation equation includes both current and expected income variables. Additionally, annuitized net worth enters the LIH model as a separate predictor variable to account for the influence of existing wealth holdings on well-being.
Kyrk (1953) suggested that the greater the discrepancy between one's resource level aspiration and actual experience, the less satisfied one becomes. In her view, individuals form expectations about their desired living standard based on personal experience and goal orientations, such that the relevant comparisons are between current income and a hypothetical, desired income level (1953). Although salient experiences for developing an income goal could include observing peers' consumption (Duesenberry also recognized that a rising standard of living is an American goal), the distinctive feature of Kyrk's idea is that consumers ultimately set and seek to meet their own income standards rather than the standards of others. Thus, Kyrk's hypothesis is unique in that it accounts for variation in ambition or indifference about material well-being that would operate quite apart from what "the Jones's" do.
This study took advantage of a unique data set from the Basic Needs Study (BNS), which obtained reports of subjective well-being in the context of an income and expenditure survey of Wisconsin residents conducted by the Institute for Research on Poverty at the University of Wisconsin in 1981 and 1982. A description of the study's aims and procedures is provided in the final report (MacDonald 1984). The BNS was funded by the Social Security Administration to provide guidance to state governments in setting standards for public assistance payments. This purpose required a comprehensive source of detailed information on households' objective economic circumstances (including complete measures of income, net worth, and expenditures), sociodemographic characteristics, and a variety of subjective measures about respondents' perceptions of their standard of living and economic needs. The BNS also replicated the questions that provide data for construction of life satisfaction variables consistent with the "social indicators" literature (Andrews and Withey 1976) and applied by home economists in the "quality of life" literature (Ackerman and Paolucci 1983).
Respondent householders were interviewed five times (one personal and four telephone interviews) over the course of 18 months (during 1981-1982). The respondent was defined as the individual with the most responsibility for the household's finances, and the household was defined as the set of people at one address related by blood, marriage, other legal arrangement, or who shared living expenses. The 1,816 persons who completed the personal interviews were originally contacted through random digit dialing within pre-specified geographic areas.(2) The response rate for the personal interviews was 67 percent. The personal interviews collected background demographic information and extensive reports on all income sources for the respondents and other household members as well as containing a complete set of questions for assets and debts. The personal interview respondents were also asked to complete three two-week diaries of all expenses during the months to which their subsequent telephone interviews referred. The diaries were intended to measure food expenditures and other frequently recurring financial outlays, such as for personal care products and gasoline. Information from the first telephone interviews was used in this study to provide less frequent expenditures, debt balances, changes in household demographics, and subjective evaluation of well-being.
The multiple contact method employed in the BNS contributed to problems of sample attrition. Two main sources of attrition were respondents who could not be located for the follow-up telephone interview and those who did not complete the expenditure diary. Two-hundred-forty personal interview respondents could not be located or refused to participate in the first telephone interview. About 80 percent of the telephone survey respondents completed the diary. Thus the final response rate, which included completion of the first personal interview, the first telephone interview, and the expenditure diary, was 46 percent of the initial sample frame. Earlier analysis revealed that sample selection bias was not a problem (Douthitt, MacDonald, and Mullis 1991).(3)
The theoretical foundation and focus here are on households that pool their resources in order to meet member needs. As the BNS sample represented all Wisconsin households, it was necessary to exclude households that may not be income poolers by restricting analysis to a sample subset. Hence the final sample used for this analysis was drawn by eliminating: single, childless individuals under 40 years of age; respondents with children over 30 years of age living at home; and households with either nonrelatives or extended family members present. This left 765 households that completed both interviews and the expenditure diaries. Table 1 provides summary sample statistics.
TABLE 1 Descriptive Statistics for Variables (n = 765) Weighted Mean Std. Dev. Family Income $20,765 14,312 Family Size 2.71 1.48 Total Monthly Expenditures $1,551 1,283 Expenditures/Median Income 1.41 1.16 Household Equivalence 1.89 3.42 Percentage of Total Sample: Percent Childless Couples 7 Two Parent Families with: Preschool Children 11 School Age Children 19 Launching Children 12 Single Parent Families with: Preschool Children 1 School Age Children 7 Empty Nest Families 23 Older Singles 20
Many scales have been created to measure psychological well-being (Bradburn and Caplovitz 1965; Campbell, Converse, and Rodgers 1976; Herzberg and Hamlin 1961; Herzberg, Mausner, and Snyderman 1959). A constructed indicator of life satisfaction provided the dependent variable, which was selected on the basis of an analysis devoted to the appropriate specification of the dependent variable (Douthitt, MacDonald, and Mullis 1991). With respect to explanatory power for understanding how family size and current income or a measure of relative income (DIFFERENCE defined below) influenced subjective well-being, this analysis demonstrated that a dependent variable constructed from many domain-specific scales was superior to the single global satisfaction question, "How do you feel about life as a whole?"
To develop this dependent variable, factor analysis of responses to 14 BNS items assessing respondent satisfaction with various aspects of their lives was conducted on Andrews and Withey's (1976) seven-point (1 = terrible; 7 = delighted) scale. Factor weights were then applied and the weighted sum divided by the number of items to construct an index of overall well-being. Because three of the 14 items related to job satisfaction, separate weights were calculated for the two groups of employed versus nonemployed respondents. Items used in the constructed index included satisfaction regarding family life, physical needs, accomplishments, government/economy, leisure activities, standard of living, housing, health, financial security, cost of living, income, job, employment remuneration, and work environment.(4)
A single common factor analysis of responses to these 14 items revealed the factoral structure among the observed domain-specific items. The eigenvalue (characteristic root) of the underlying factor for the employed subsample was 4.02 and it accounted for about 29 percent of the covariance among the 14 indicators. The eigenvalue for the nonemployed subsample was 3.62, accounting for 33 percent of the covariance among 11 indicators. All of the observed items were positively related to the underlying factor, overall life satisfaction, and each carried a factor weight greater than 0.3. For both the employed and nonemployed respondents satisfaction with standard of living, financial security, and income loaded the highest. Because many of the nonemployed respondents were retired, satisfaction with health and physical needs weighted heavier than for their employed counterparts.
MEASURES OF ECONOMIC WELL-BEING
Three measures of economic well-being were constructed to test the relationship between psychological and economic well-being, corresponding to the life cycle, relative consumption, and resource deficit theories.
Life Cycle Income Hypothesis
The life cycle income hypothesis (LIH) was operationalized by measuring respondent's current and expected income as well as net worth. Consistent with the life cycle concept of economic well-being, these variables indicated the consumer's perceived ability to maintain a relatively level pattern of consumption throughout life.
Current income (INCOME) was measured as the sum of all sources of gross, before tax family income for the year 1980. Consistent with the speculation that the marginal utility of income may diminish, the natural log metric rather than absolute dollars was used to measure the effects of income on life satisfaction. This corresponds to the conjecture that additions to income produce less utility than prior increments. Though part of the economists' lore, the hypothesis cannot be formally tested because the marginal utility of income depends on unobservable preferences (Deaton and Muellbauer 1980).
Expected income (EXPECTED INCOME) was calculated for each household by (1) regressing current total family income on respondent education, respondent occupation, and youngest adult's age and age squared; (2) using estimated parameters from step (1) to generate a life cycle income stream between ages over periods T + 1 to D (i.e., a stream for the remaining life expectancy of the youngest adult, D); (3) summing that stream; and (4) dividing by remaining life expectancy of youngest adult in the household (D - T + 1). An annual annuitized value of net worth (ANNUITIZED NET WORTH) was calculated by subtracting liabilities from assets and dividing by the remaining life expectancy of the youngest adult in the household.
Relative Income Hypothesis
Duesenberry's relative income hypothesis (RIH) was operationalized by including not only current income (INCOME) but also a measure of relative income. Current income was measured as described in the previous LIH section. "Relative" income was defined as the difference between the respondent's actual total monthly expenditures and the sample mean monthly expenditures for goods and services. This captured the "relative" nature of economic well-being or the extent to which families are keeping up with the average family's (Jones's) consumption. As an alternative to using the sample mean expenditure as a common reference category, the expenditure mean within ten-year age groups (20-29, 30-39, etc.) was used to create another measure of relative income, which behaved similarly to the one for the sample mean and did not alter the other regression coefficients.(5)
Resource Deficit Hypothesis
Kyrk's resource deficit hypothesis (RDH) was operationalized by examining the BNS question, "How much income would make you feel terrible?" (ACTUAL MINIMUM INCOME), and by subtracting that value from current family income (previously defined).(6) This was one of seven questions which asked respondents to define an income amount that corresponded with the adjectives "terrible," "unhappy," "mostly dissatisfied," "mixed," "mostly satisfied," "pleased," or 'delighted."
Life Cycle Categories
Wilensky (1981) and Estes and Wilensky (1978) have conducted extensive research to analyze the relationship between stage of the life cycle and sense of well-being. Wilensky (1981) developed the concept of "life cycle squeeze" which represents stages during which people experience minimal job satisfaction, lowest participation in community life, greatest financial and family burdens, and greatest psychological tension. Table 2 defines the life cycle stage dummy variables used to control for the influence of these factors and modified to incorporate single parents status as an increasingly important life cycle stage. There may be other relevant aspects of the life cycle, such as the distinction between first marriage and remarriage or never married and ever married. The need to have a sufficient number of cases in each category of the life cycle variable accounts for the abbreviated categorical specification in Table 2.
TABLE 2 Variable Definitions for Analysis of Subjective Well-Being INCOME Total family gross income for 1980. Includes earned and unearned sources. EXPECTED INCOME Average annual income calculated by predicting a life cycle stream of income and dividing by the remaining years of life expectancy. DIFFERENCE Monthly household expenditures divided by the sample mean and monthly expenditures. HH EQUIVALENCE Adult equivalency household size where a 19 year old male is the standard consumer ACTUAL MINIMUM INCOME Difference between current income and amount of income that respondents report could make them feel "terrible." ANNUITIZED NET WORTH Total value of family assets net total family debt annuitized on an annual basis over the remainder of the life cycle. Life Cycle Variables MARRIED: YOUNG CHILDLESS One is respondent is 40 years or younger and has no children present; zero otherwise. PRESCHOOL CHILDREN One if youngest child present is less than 6 years of age; zero otherwise. SCHOOL AGE CHILDREN One if the youngest child present is 6 to 7 years of age; zero otherwise. LAUNCHING One if the youngest child present is 18 years of age or older; zero otherwise. EMPTY NEST One if respondent is 0 years or older and has no children present; zero otherwise. SINGLE: PRESCHOOL CHILDREN One if youngest child present is less than 6 years of age; zero otherwise. OTHER CHILDREN One if youngest child present is over age 6; zero otherwise. SURVIVOR One if over age 40 and has no children present; zero otherwise.
These life cycle dummies were included in the relative income and resource deficit hypotheses models. However, they did not enter the life cycle income hypothesis model. Including life cycle stage dummies would be redundant from the perspective of implementing Modigliani's hypothesis, which is parsimonious in that it relies primarily on net worth and current versus expected incomes without reference to a particular age or life cycle category.(7) Additionally, tests revealed that the respondent's age was significantly correlated with six of the seven life cycle stage categories (the two highest correlation coefficients were 0.51 and 0.52, respectively). Thus age was also not included in the life cycle income regression.
Adult-Equivalent Consumption Needs
In addition to measures of economic well-being and life cycle stage effects, it was necessary to control for family needs in testing the relationship between the life cycle income hypothesis and life satisfaction. In previous studies of well-being, often such measures were either omitted or a gross indicator like family size was used. For this study, household equivalence scales were constructed (HH EQUIVALENCE) to control for such effects.
The particular household equivalence measure used in this analysis was developed by Buse and Salathe (1978) and modified by Tedford, Capps, and Havlicek (TCH) (1986). The primary advantage of the Buse and Salathe equivalence scale calculation is that the scale is expressed as a continuous rather than discrete function of age. TCH's contribution involved respecification of the scaling functions in a manner consistent with the human development literature by augmenting stages first identified by Levinson et al. (1978). Like Buse and Salathe, TCH incorporates cubic spline functions to calculate the continuous adult scale functions. In total 16 weighted sum variables were calculated as a function of household age and gender characteristics and regressed on total monthly expenditures for food to derive the household equivalence measure. Details of this procedure are available from the authors upon request.
The household equivalence measure was not included as a control in either the RIH or RDH models. It was not included in the RIH model because its construction is based on a life cycle theory and thus is highly correlated with the life cycle stage indicator variables. Life cycle variables in the RIH model capture both life cycle and need effects. In the resource deficit model, no control for family needs was required, as respondents implicitly make such assessments in expressing what level of income makes them feel terrible.
All three models were estimated using a weighted least squares regression analysis. Results are presented in Table 3. All model specifications explained a significant amount of variance (about 12 to 15 percent) in the dependent variable, overall life satisfaction. The explained variance of a simple model including only current income and family size was about seven percent (Douthitt, MacDonald, and Mullis 1991).
Life Cycle Income Hypothesis Results
Results of the LIH model indicate that both current and expected family income as well as the household equivalence variables are significantly related to life satisfaction. Current income is positively related to life satisfaction as expected. However, expected future income is negatively related to current life satisfaction. Holding the effects of current income constant, this implies that those with greater expected future or permanent incomes report lower current levels of satisfaction. One explanation of this puzzling result may be that capital market imperfections impede borrowing such that the reduced current satisfaction reported by respondents with higher permanent incomes reflects their impatience. Thus, respondents who cannot currently capitalize on their higher expected future incomes may be frustrated by an inability to consume at the appropriate "life cycle" level. Those with lower permanent incomes may be currently at their earnings peak and thus more apt to express greater satisfaction with their current circumstances, because they suffer less from an inability to borrow. The result may also reflect the sentiments of TABULAR DATA OMITTED permanently low income consumers with little expectation of capitalizing on future earnings.(8)
The household equivalence parameter is of the expected sign, indicating that increasing family size by one standard unit (in this case the equivalent of a 19 year old male), holding income constant, results in a diminished level of perceived life satisfaction. The annuitized net worth variable proves insignificant, although of the predicted sign. Specifications that entered assets and debts separately in the same model also produce insignificant coefficients for those components of net worth.
Relative Income Hypothesis Results
Of the three models, the RIH model explains the largest percentage of variance in the dependent variable. Results again support the positive and significant relationship between current family income and overall life satisfaction. The Deusenberry relative income hypothesis variable, DIFFERENCE, indicates that as one's expenditures exceed those of an average household (holding current income constant) life satisfaction is diminished. To the extent that this variable also captures whether the household's savings rate (or perhaps more importantly dissavings rate) falls short of that of a reference household, it would imply that such behavior leads to diminished overall life satisfaction. Hence under this interpretation, the RIH results conform exactly to what Duesenberry suggested about savings as an influence on consumers' utility when preferences are interdependent.
In addition, life cycle variables are highly related to reported life satisfaction. In this analysis the reference category was young childless couples (YOUNG CHILDLESS). Results indicate that all other groups with the exception of couples at the empty nest stage of the life cycle report feeling lower life satisfaction.
By virtue of the fact that financial resources are carefully accounted for, the life cycle parameters afford the ability to more extensively analyze the effects of other stressors and needs associated with particular life cycle stages. Wilensky and others have interpreted similar patterns as evidence for a "life cycle squeeze" experienced in midlife and primarily by parents. Evidently after accounting for the greater financial burdens, the remaining difficulties of job satisfaction, reduced community life participation, and related psychological tensions still exert important independent influences on subjective well-being. Additionally, note that the life cycle results for single parent status are much more negative than for comparable categories of married couples.
Resource Deficit Hypothesis Results
The RDH results explain the second greatest percentage of variance in the dependent variable of the three models. As expected, the greater the difference between respondents' current incomes and the levels of income that they reported causes them to feel terrible (ACTUAL MINIMUM INCOME) the higher their reported current life satisfaction. This result is consistent with Kyrk's hypothesis that consumer satisfaction is related to perceptions of an adequate standard of living (1953).
Similar to the RIH findings, family life cycle indicator variables are also significant in explaining life satisfaction. The RDH results differ however, in that survivor households do not report being significantly less happy than young childless couples, controlling for current income adequacy. One difference between the two models' specifications affecting life cycle variable interpretation is that the life cycle variables in the RIH model capture both need and stressor effects while only stressor effects are captured in the RDH model. Thus, once basic needs are accounted for, single survivors are as happy as young childless couples.
Generally, all three models of economic well-being perform favorably in explaining consumer utility or psychological well-being. The relative income hypothesis model is the simplest and most powerful of the three. The theoretical specification of the relative income model addresses the question of whether keeping up with the Jones's really contributes to or detracts from one's overall life satisfaction. Results indicate that deviation from average peer expenditure patterns is important, but that spending less than average, holding income constant, improves perceived life satisfaction. Although this seems counter-intuitive when focusing on consumption alone, Duesenberry's insight that tensions about the need to save affect utility and behavior resolves the apparent puzzle. Furthermore, because savings behavior occurs over many periods the apparent success of the relative income hypothesis provides strong evidence to support the view that subjective well-being is not only a function of current economic status but of longer term resource considerations as well.
Two additional important conclusions from this work can be drawn. First, a careful specification of economic variables in models explaining family life satisfaction has important payoffs in terms of explained variance. However, which economic theory of utility maximization is used may not be as critical. The more complexly specified life cycle income hypothesis model does not perform any better than the other models with variables that are much easier to derive (but for Duesenberry's hypothesis this assumes that expenditure data are available). Secondly, when economic factors are included in life satisfaction models, the long standing a priori economic postulate that increased resources lead to higher levels of satisfaction is supported.
Maurice MacDonald and Robin A. Douthitt are Professors, Department of Consumer Science, University of Wisconsin, Madison.
Funding for this research was provided by the National Institutes for Mental Health (Grant #MH41783) and the School of Family Resources and Consumer Sciences. An earlier version of this paper was presented at the 1990 annual conference of the American Council on Consumer Interests in New Orleans, LA. The authors would like to acknowledge the generous help of anonymous referees, thoughtful comments from Charles Hennon, and the expert programming assistance of Susan Bruns.
The hypothesis of utility maximization (and perfect markets) has, all by itself, one very powerful implication--the resources that a representative consumer allocates at any age . . . will depend only on his life resources . . . and not at all on income accruing currently (Modigliani 1986, 299).
From the viewpoint of preference theory or marginal utility theory, human desires are desires for specific goods; but nothing is said about how these desires arise or how they are changed. That, however, is the essence of the consumption problem when preferences are interdependent (Duesenberry 1949, 19).
Everyone probably has, more or less consciously formulated, an ideal standard of living, a level toward which he moves as income and other opportunities permit; he has also . . . a standard that he insists upon maintaining. To attain the first would be a highly desirable state of economic well-being, to attain the second is essential, and to fall below it is intolerable (Kyrk 1953, 374).
1 Transitory shocks to current income cause departures from the normal or permanent income path. As an alternative to the LIH model one could rely on the permanent income hypothesis, which specifies that consumption is controlled by normal or "permanent" income rather than current income (Friedman 1957). Thus current consumption could be used to gauge the short term level of permanent income for explaining life satisfaction in the current period. However it remains an empirical issue whether respondents' life satisfaction assessments incorporate their expectations about subsequent period incomes.
2 The entire BNS sample frame consisted of 2,718 Wisconsin households. This included 1,220 randomly selected households, supplemented by telephone screening procedures to oversample: 330 households currently receiving Aid to Families with Dependent Children; 226 headed by a female without a male adult in the household; 485 headed by a person who was 65 years or older; and 457 households with a monthly income below 144 percent of the 1980 food stamp income eligibility limit. Sample weights were used in the analyses to adjust for over-sampling to permit inferences for a representative sample of Wisconsin households.
3 The first stage of the research analyzed the importance of variable specification in using economic variables to explain perceived overall psychological, economic, and noneconomic well-being. The Heckman procedure was applied to correct for sample selection bias, hypothesizing that family income, region of the state (urban versus rural counties), respondent's age, education, and gender would affect the probability that a personal interview respondent would have provided the other necessary segments of the survey. None of these variables was significant. The variables and the inverse of the Mills' ratio obtained from this analysis did not affect a variety of regressions of current income and family size on subjective well-being.
4 Because part of the research effort is devoted to understanding the explanatory power of objective economic variables compared to that of other independent variables a criticism of this dependent variable has been that it includes components from economic domains, hence potentially "biasing" the analysis in favor of the economic variables. This issue is not relevant here, as the purpose of this study is to compare results within a list of objective economic indicators. However, note that the earlier results with respect to the general explanatory power of objective economic variables are not sensitive to whether the dependent variable includes the items for "cost of living," "income," "job," and "employment remuneration" (Douthitt, MacDonald, and Mullis 1991).
5 Certainly other alternatives are consistent with Duesenberry's hypothesis, such as the mean expenditure of persons in the "next highest" income group or other social reference groups (such as one's siblings, birth cohort, educational level, or occupational category). For each of these reference groups one has to account for the independent effects of the variable defining the reference group which complicates the question at issue in this paper.
6 The researchers considered using the "mostly dissatisfied" and "unhappy" statements as the reference income level for measuring the deficit at interest but decided to focus on the "terrible" level to increase the chances of finding effects for the more extreme discrepancy that would be associated with "terrible." This decision supports the attempt to understand the relative explanatory power of the competing hypotheses, and seems to give the resource deficit approach the best opportunity to explain variation in well-being. However it could be that the respondents' sense of well-being is dependent on their perception of potential income losses that are most likely, such that "mostly dissatisfied" and "unhappy" would be more salient for them--a possibility that was not explored.
7 Including life cycle stage variables may increase the explanatory power of the model, and an appropriate set of demographically defined life cycle stage categories could capture the effects of associated income and asset variations on well-being in a manner that is consistent with Modigliani's theory. These issues are the subject for another analysis. Work in progress includes a more thorough analysis of alternative specifications of life cycle variables, in which the role of life stage as the main determinant of well-being from a social-demographic perspective is explored.
8 Modigliani considered this type of imperfection in credit markets as one of five important complications for the empirical application of his original, simplified savings hypothesis. (The others include the role of interest rate, family size and the life cycle of earnings, length of working and retired life, and myopia.) Specifically, capital market imperfection may "prevent households from borrowing as much as would be required to carry out the unconstrained optimum consumption plan" (1986, 305).
Ackerman, N. M. and B. Paolucci (1983), "Objective and Subjective Income Adequacy," Social Indicators Research, 12: 25-48.
Allardt, E. (1977), "On the Relationship Between Objective and Subjective Predicaments," Research Reports, 16, University of Helsinki: Research Group for Comparative Sociology: 1-22.
Andrews, F. M. and A. C. McKennell (1980), "Measures of Self-reported Well-being: Their Affective, Cognitive, and Other Components," Social Indicators Research, 8: 127-155.
Andrews, F. M. and S. B. Withey (1976), Social Indicators of Well-being: Americans' Perceptions of Life Quality, New York and London: Plenum.
Bradburn, N. M. and D. Caplovitz (1965), Reports on Happiness: A Pilot Study of Behavior Related to Mental Health, Chicago: Aldine.
Buse, R. C. and L. E. Salathe (1978), "Adult Equivalence Scales: An Alternative Approach," American Journal of Agricultural Economics, 60: 460-468.
Campbell, A. (1976), "Subjective Measures of Well-being," American Psychologist (February): 117-124.
Campbell, A., P. E. Converse, and W. E. Rodgers (1976), The Quality of American Life: Perceptions, Evaluations, and Satisfactions, New York: Russell Sage Foundation.
Deaton, A. and J. Muellbauer (1980), Economics and Consumer Behavior, New York: Cambridge University Press.
Dohrenwend, B. S. (1973), "Life Events as Stressors: A Methodological Inquiry," Journal of Health and Social Behavior, 14: 167-175.
Douthitt, R. A., M. MacDonald, and R. Mullis (1991), "The Relationship Between Measures of Subjective and Economic Well-Being: A New Look," Social Indicators Research, 25: 91-106.
Duesenberry, J. S. (1949), Income, Saving, and the Theory of Consumer Behavior, Cambridge, MA: Harvard University Press.
Estes, R. and H. L. Wilensky (1978), "Life Cycle Squeeze and the Morale Curve," Social Problems, 25: 227-292.
Fernandez, R. and J. C. Kulik (1981), "A Multilevel Model of Life Satisfaction: Effects of Individual Characteristics and Neighborhood Composition," American Sociological Review, 46: 840-850.
Friedman, M. (1957), A Theory of the Consumption Function, Princeton, NJ: Princeton University Press.
Gray, L. C., H. F. Goldsmith, B. B. Livieratos, and H. J. Dupuy (1983), "Individual and Contextual Social Status Contributions to Psychological Well-being," Sociology and Social Research, 68: 78-95.
Hall, J. (1976), "Subjective Measures of Quality of Life in Britain 1971 to 1975: Some Developments and Trends," Social Trends, 7: 47-60.
Haring, M. J., W. A. Stack, and M. A. Okun (1984), "A Research Synthesis of Gender and Social Class on Subjective Well-being," Human Relations, 37: 645-657.
Herzberg, F. and R. M. Hamlin (1961), "A Motivation-hygiene Concept of Mental Health," Mental Hygiene, 45: 394-401.
Herzberg, F., B. Mausner, and B. Snyderman (1959), The Motivation to Work, New York: John Wiley and Sons.
Keith, P. (1985), "Changing Patterns and Life Satisfaction," in Later Life Transitions: Older Males in Rural America, E. A. Powers, W. J. Goody, and P. M. Keith (eds.), Bingham, MA: Kluwer Academic Publishers: 137-156.
Kessler, R. C. (1982), "A Disaggregation of the Relationship Between Socioeconomic Status and Psychological Distress," American Sociological Review, 47: 752-764.
Kyrk, H. (1953), The Family in the American Economy, Chicago: The University of Chicago Press.
Levinson, D. J., C. N. Darrow, E. B. Klein, M. H. Levinson, and B. McKee (1978), The Seasons of a Man's Life, New York: Ballantine Books.
Levy, F. (1987), Dollars and Dreams: The Changing American Income Distribution, New York: Russell Sage Foundation.
Link, B. (1982), "Mental Patient Status, Work, and Income: An Examination of a Psychiatric Label," American Sociological Review, 47: 202-215.
MacDonald, M. (1984), "Evaluating Alternative Approaches to Measuring Basic Needs," Wisconsin Basic Needs Study Final Report to the Social Security Administration, Institute for Research on Poverty, University of Wisconsin-Madison (July).
Modigliani, F. (1986), "Life Cycle, Individual Thrift, and the Wealth of Nations," American Economic Review, 76 (3, June): 297-313.
Modigliani, F. and R. Brumberg (1954), "Utility Analysis and the Consumption Function: An Interpretation of Cross-section Data," in Post-Keynesian Economics, K. K. Kurihara (ed.), New Brunswick, NJ: Rutgers University Press: 338-436.
Mutran, E. and D. C. Reltzes (1984), "Intergenerational Support Activities and Well-being Among the Elderly: A Convergence of Exchange and Symbolic Interaction Perspectives," American Sociological Review, 49: 117-130.
Murphy, P. E. and W. M. Staples (1979), "A Modernized Family Life Cycle," Journal of Consumer Research, 6: 12-22.
Stampfl, R. (1978), "The Consumer Life Cycle," The Journal of Consumer Affairs, 12 (2, Winter): 209-219.
Tedford, J. R., O. Capps, Jr., and J. Havlicek, Jr. (1986), "Adult Equivalent Scales Once More--A Developmental Approach," American Journal of Agricultural Economics, 68: 322-333.
Wilensky, H. L. (1981), "Family Life Cycle, Work, and the Quality of Life: Reflections on the Roots of Happiness, Despair, and Indifference in Modern Society," in Working Life, B. Gardell and G. Johansson (eds.), New York: John Wiley & Sons Ltd.: 235-265.
|Printer friendly Cite/link Email Feedback|
|Author:||MacDonald, Maurice; Douthitt, Robin A.|
|Publication:||Journal of Consumer Affairs|
|Date:||Dec 22, 1992|
|Previous Article:||Consumer access to health care: basic right 21st century challenge.|
|Next Article:||Factors affecting borrower choice between fixed and adjustable rate mortgages.|