International Review of Economics

, Volume 57, Issue 2, pp 163–176

Does consumption buy happiness? Evidence from the United States

Authors

    • University of Wisconsin-Madison
    • IZA
    • National Bureau of Economic Research
  • Ariel Kalil
    • University of Chicago
Article

DOI: 10.1007/s12232-010-0093-6

Cite this article as:
DeLeire, T. & Kalil, A. Int Rev Econ (2010) 57: 163. doi:10.1007/s12232-010-0093-6

Abstract

We examine the association between various components of consumption expenditure and happiness in the Health and Retirement Study (HRS), a nationally representative sample of older Americans. We find that only one component of consumption is positively related to happiness—leisure consumption. In contrast, consumption of durables, charity, personal care, food, health care, vehicles, and housing are not significantly associated with happiness. Second, we find that leisure consumption is associated with higher levels of happiness partially through its effect on social connectedness, as indexed by measures of loneliness and embeddedness in social networks. On one hand, these results counter the conventional wisdom that “material goods can’t buy happiness.” One the other hand, they underscore the importance of social goods and social connectedness in the production of happiness.

Keywords

HappinessLeisureConsumption

JEL Classification

D12I31

1 Introduction

The term “subjective well-being” is often used interchangeably with the more colloquial term “happiness.”1 This construct refers to people’s cognitive and affective evaluations of their lives (Diener 2000). In principle, it is the most democratic of evaluations, as it allows each individual to decide for him or herself whether they have had a good and worthwhile life. Among valued life goals, it is the highest-ranked across many countries and among many different demographic groups (Diener 2000).

Research suggests that although happiness has a strong genetic component, about 50% of the differences between people in their life happiness can be attributed to external factors (Weiss et al. 2008). While psychologists have long been interested in the external factors that correlate with subjective well-being, only relatively recently has this topic captured the attention of economists. It is perhaps not surprising that a major emphasis in economics studies is on the links between income and subjective well-being (for example, Clark and Lelkes 2005; Dehejia et al. 2007; Easterlin 1995; Layard 2005).

Our paper examines the question of whether and how consumption is related to happiness using newly available data from a large nationally representative U.S. data source. The paper is structured as follows: Sect. 2 briefly reviews the existing evidence on income and subjective well-being and outlines the main contributions of our analysis. Section 3 outlines the data and our analytic model. The results of our estimation are reported in Sect. 4, and Sect. 5 concludes.

2 Income and subjective well-being

In this section, we discuss the relevant literature on income and happiness and the contributions of our study.

2.1 Literature review

How much does money matter to our happiness? This question has been asked with respect to trends at the aggregate level (i.e., the mean level of subjective well-being of the society) and the individual level. Because the present paper is concerned with the latter, we will focus on that literature here (see Diener 2000; Diener and Diener 1995; and Stevenson and Wolfers 2008, for discussion of aggregate trends in happiness).

The extent to which income contributes to happiness is a matter of some debate. Though cross-sectional micro-data analysis suggests positive and statistically significant correlations between life satisfaction and income, the magnitude of these associations is generally small (Powdthavee 2010) and has led scholars to conclude that the relationship between money and happiness is weak at best (Layard 2005). Indeed, differences in the personal income of individuals explain less of the difference in reported well-being than a range of other factors, such as employment, family relationships, health, education, and income inequality (Di Tella and MacCulloch 2006; Winkelmann and Winkelmann 1998).

This question may be more fruitfully understood, however, if it were re-phrased as “what is it (if anything) about money that makes us happy?” In other words, turning our attention from the level of individuals’ household income (the key variable of interest in most studies) to measures of how individuals spend the money they have (i.e., the composition of spending) provides an opportunity for testing more specific hypotheses about why money may (or may not) matter to individual life satisfaction. We know of no existing studies that have addressed the question in this fashion, but we offer some guiding hypotheses here.

First, increased consumption of goods may increase well-being by reducing material hardship or making life easier, thereby enhancing happiness. This idea, for example, might be reflected in a positive association between the purchase of durable goods, such as a washing machine, or spending on food or housing, and happiness.

Second, consumption of certain “conspicuous” goods may increase happiness by increasing status (Veblen 1899; Charles et al. 2009). Clothing, jewelry, and automobiles have been characterized as “status” goods or “visible” goods and may be related to happiness in this fashion as may spending on expensive vacations or the trappings of expensive hobbies or athletic pursuits.

Third, a hypothesis that we are especially interested in is that income may be linked to happiness through its effect on social relationships. There is substantial evidence that social relationships bring happiness to individuals (Diener and Biswas-Diener 2002; Putnam 2000). This suggests that income could “buy” happiness if it enhances social connectedness or vitality. For instance, spending on leisure might enhance life satisfaction because leisure is often enjoyed in the company of friends, relatives, and neighbors. Some forms of charitable activities may be associated with social activities and could increase happiness through this channel. Conversely, the mere accumulation of material goods may fail to increase life satisfaction if it does nothing to strengthen social bonds. Indeed, one recent explanation for the lack of aggregate increase in subjective well-being even as GDP has increased is that any potential positive impact on subjective well-being has been offset by the weakening over time in social bonds and a rise over time in materialism as described by Putnam (2000; see Pugno 2008 for more on this point). There is also evidence that people who are more oriented toward the consumption of market goods for the sake of materialism (and, conversely, less oriented toward relational goods) report lower levels of subjective well-being (Kasser 2002).

Of course, we do not know which goods have a social component and which do not. Because of this, we implement a method to indirectly determine whether any association between the consumption of a good and happiness is due to increased social connectedness, as described below.

2.2 Contributions of the present study

Our study makes several important contributions to a literature that still has many unanswered questions. First, we use individual data from a large nationally representative data set (the U.S. Health and Retirement Survey or HRS) that, in addition to tracking the health and employment of a cohort of older Americans, provides detailed measures of expenditure on a wide variety of goods for this cohort. Economists have long argued that consumption data are a superior measure of permanent income and overall material well-being than are income data (see, for example, Meyer and Sullivan 2003). Moreover, expenditure data are arguably superior to measures of income for questions such as the ones posed here because they capture whether and in what ways individuals choose to spend the income they have. The conventional wisdom that says “money can’t buy you happiness” is often intended to mean that “spending on material goods can’t buy you happiness” but, in the broader literature, studies commonly pose the empirical question of whether “income is associated with happiness.” Our approach addresses a key underlying question in this research area by investigating what it is that money allows us to acquire (as opposed to other uses of income such as saving or paying down debt) and whether the acquisition of specific goods or services makes us happy.

Second, our data contain a high-quality, multi-item index of subjective well-being that has superior measurement properties compared to single-item measures of happiness that are often used in this literature. Studies of subjective well-being rarely take a comprehensive set of measures and instead often use generic measures such as “all things considered, how happy are you” rather than constructing indicators that target positive and negative emotions (Diener and Seligman 2004). Subjective happiness appears to vary according to the time of day and season (Layard, 2005), phases of an economic cycle, population age-profile, and differences between expectations and outcomes. Thus, the timing of information gathering on happiness status and its interpretation (permanent or transient effects) is an important complicating factor in happiness measurement. Our measure of life satisfaction, which is gathered from a sample of older adults, poses questions in a way that prompts respondents to take the long view of their life and assess how things have turned out for them, rather than linking point-in-time (and likely ephemeral) states of happiness to specific events, dates, or purchases.

Third, our data on subjective well-being and our data on expenditure were collected prospectively on different survey occasions (during the same year), and thus avoid the recall and other biases associated with methods that ask respondents to describe purchases they made in the past and then indicate how happy they thought those purchases made them feel.

Fourth, our data contain measures of social interactions and correlated psychosocial and personality characteristics (e.g., depressive symptoms, loneliness) that allow us to test the key hypothesis that spending can increase happiness via the mechanism of strengthened social bonds. We also include controls for personality characteristics (e.g., neuroticism) that may affect expenditure and happiness. Including such controls allows us to rule out some potential sources of unobserved heterogeneity. We know of no such study that uses as high-quality data that answers the question we pose here.

In sum, the specific research questions addressed by the study are as follows:
  1. 1.

    Which, if any, components of consumption expenditure are associated with higher levels of subjective well-being?

     
  2. 2.

    To the extent that consumption expenditure is associated with happiness, is it the consumption of material goods itself or the social connections that come with certain forms of consumption that lead to happiness?

     

As we discuss below, our findings suggest that spending on leisure goods and activities (vacations, entertainment, sports, and leisure equipment) is associated with higher levels of happiness. By contrast, spending on other types of consumption (food, utilities, health care, etc.) is not associated with increased happiness. Our findings also suggest that leisure spending increases happiness in part through actual or perceived social connections that are linked to leisure spending, rather than the consumption of the material goods themselves. We thus conclude by rejecting the conventional wisdom that “material goods can’t buy happiness,” and instead suggest a potential role for social connections and status in the production of subjective well-being.

3 Data and methods

This section provides a detailed description of the data set we use for the empirical analysis, the particular measures we use from those data, and the empirical methods we employ to address our key research questions.

3.1 Data

We use the Health and Retirement Study (HRS), a data effort funded primarily by the National Institute on Aging and collected by the University of Michigan, as our primary data source for this project. The HRS is designed by an interdisciplinary group of investigators from economics, demography, medicine, psychology, sociology, and survey methods from the University of Michigan and from universities and research institutions around the country (Willis 2006). Comprised of approximately 20,000 individuals, the HRS provides a nationally representative portrait of the United States population over the age of 50. It is an ongoing longitudinal study that began in 1992 and has tracked respondents approximately every 2 years subsequently. Eight waves of data are currently available; 2006 is the most recent year for which data are available. We use data on the cohort of individuals born between the years 1931–1941; these individuals are referred to as the initial HRS cohort.

The HRS contains a core survey that is asked of all participants in every round. The core survey includes detailed questions on employment, income, health, family structure, and wealth. However, the HRS project creates a data system that extends beyond the core survey waves. In particular, the HRS supplements the core survey with data from separate “modules” that collect data on a wide variety of other measures. Generally these modules are asked in one wave for a subsample of respondents.

The 2006 wave of the HRS included one of these new “modules” that focused on respondents’ life satisfaction and a wide range of related measures of psychological well-being. This module (the “Leave-Behind Participant Life Style Questionnaire”; see Clarke et al. (2007) for further information) was administered at random to the half sample of participants who completed a face-to-face 2006 core survey. Specifically, there were 8,566 eligible respondents for the 2006 Leave-Behind Questionnaire. The completion rate was about 86%. Consequently, 7,365 respondents completed the questionnaire (we exclude 97 cases that were completed by someone other than the assigned respondent).

The other main source of data we rely on is the Consumption and Activities Mail Survey (CAMS). The CAMS is a self-administrated questionnaire mailed to HRS respondents that asks about their household consumption and expenditure patterns during last year. As of 2008, a total of four waves of CAMS (i.e., 2001, 2003, 2005, and 2007) are available. Our analysis uses the 2007 CAMS, which assesses respondents’ consumption and expenditures in 2006 (we use this wave of CAMS to match the year in which the psychosocial data were collected). In the fall of 2007, a total of 5,209 CAMS questionnaires were mailed to randomly selected HRS respondents. Of these, 3,738 were returned.

The analytic sample for the present paper reflects the intersection of respondents in 2006 who completed both the 2007 CAMS and the 2006 Leave-Behind Questionnaire. This represents 1,830 HRS respondents. A fair number of these respondents have missing consumption data for some, but not all, of the consumption categories. Only 860 respondents have no missing consumption data. Because of this, we impute missing values for consumption using mean imputation as well as multiple imputation methods (Rubin 1987). These analyses yielded generally similar results. However, because the multiple imputation method rests on the relatively strong assumption that all missing data are missing (conditionally) at random, we rely on the results from the mean imputation method as our primary ones.2 All of these analyses also include a dummy variable indicating whether the value for the consumption data was imputed. Accounting for the small amount of missing values on other independent and dependent variables, our final analytic sample is 1,733.

At this time, longitudinal analysis of the relationship between consumption, happiness, and the psychosocial measures is not possible using the HRS, though it should be possible as more waves of the LBQ become available and as its coverage increases.

3.2 Measures

The key measures from the 2006 HRS psychosocial survey and core survey are described below.

3.2.1 Outcome: subjective well-being

Subjective well-being (or happiness)—the key outcome measure of interest—is constructed using the satisfaction with life scale, a five component measure administered in the psychosocial questionnaire. Specifically, respondents are asked to rate how much they agree or disagree with the following five statements: “In most ways my life is close to ideal”; “The conditions of my life are excellent”; “I am satisfied with my life”; “So far, I have gotten the important things I want in life”; and “If I could live my life again, I would change almost nothing.” The scores from these five items are averaged to create the “satisfaction with life scale”, one of the most frequently used measures of subjective well-being. Measuring life satisfaction in this manner follows Diener and Biswas-Diener (2002); the measure exhibits good psychometric properties (alpha of 0.89 in the HRS data).

3.2.2 Independent variables: consumption expenditure

We examine consumption expenditure for 9 categories of consumption. The unit of consumption is the annual expenditure (in thousands of dollars) in 2006. The consumption categories we examine are as follows:

Leisure Trips and vacations, tickets to movies, sports events, and performing arts, sports (including gym membership and exercise equipment), hobbies, and leisure equipment.

Durables Refrigerators, washing machine/dryer, dishwasher, televisions, and computers. Purchases of automobiles are included in the category vehicles, described below.

Charity and gifts Contributions to religious, political, educational, and charitable organizations, cash or gifts to family and friends outside the household.

Personal care and clothing Housekeeping supplies, housekeeping, dry cleaning, and laundry services, gardening and yard supplies, gardening and yard services, clothing and apparel, and personal care products and services.

Health care Health care services, prescription and non-prescription medications, medical supplies, and health insurance.

Food in Food and drinks, including alcohol, purchased in grocery or other stores.

Food out Dining and/or drinking in restaurants, cafes, and diners, including takeout food.

Utilities and housing Mortgage, rent, electricity, water, heating fuel for the home, telephone, cable, internet, homeowner’s or renter’s insurance, home repairs and maintenance, household furnishing and equipment.

Vehicles Vehicle purchases and all the major categories of the flow value of consumption from vehicles (following Meyer and Sullivan 2003), vehicle insurance, vehicle maintenance, and car payments (interest and principal).

3.2.3 Mediating variables: psychological well-being and social connectedness

The 2006 psychosocial questionnaire also includes key measures of psychological factors that can reflect in part the social aspect of consumption. These include depressive symptoms, loneliness, and a measure of social interaction:

Depressive symptoms We use the Center for Epidemiologic Studies Depression (CES-D) Scale, a 20-item scale that was developed for use in studies of the epidemiology of depressive symptomatology in the general population (Radloff 1977).

Loneliness Respondents are asked to answer the following three questions: “How often do you feel you lack companionship?”; “How often do you feel left out?”; and “How often do you feel isolated from others?” An index is constructed from the average of these three measures (alpha = 0.82).

Social interaction This measure captures the degree of embeddedness in social networks (clubs, groups, etc.) in terms of the frequency of social interaction within this network. Specifically, this question asks, “Not including attendance at religious services, how often do you attend meetings or programs of groups, clubs, or organizations that you belong to?” The question is coded such that 5 = more than once a week, 4 = once a week, 3 = 2 or 3 times a month, 2 = about once a month, 1 = less than once a month.

Table 1 reports summary statistics on all study variables.
Table 1

Descriptive statistics (N = 1,733)

 

Mean

Standard deviation

Range

Happiness

4.30

1.24

[1, 6]

Age

67.6

10.5

[31, 101]

Female

0.620

0.485

[0, 1]

Hispanic

0.062

0.241

[0, 1]

Non-White

0.155

0.362

[0, 1]

Not married

0.460

0.499

[0, 1]

Household income

64.05

152.26

[0, 5039.78]

Household wealth

588.20

2956.20

[−354.67, 81828.09]

Less than high school

0.176

0.145

[0, 1]

GED

0.044

0.0205

[0, 1]

High school graduate

0.317

0.466

[0, 1]

Some college

0.250

0.433

[0, 1]

College and above

0.213

0.409

[0, 1]

Neuroticism

2.087

0.590

[1, 4]

Agreeableness

3.528

0.470

[1, 4]

Extravertedness

3.198

0.546

[1, 4]

Conscientiousness

3.346

0.456

[1, 4]

Openness

2.933

0.537

[1, 4]

Depression (CESD)

1.457

1.958

[0, 8]

Loneliness

1.516

0.556

[1, 3]

Social interaction

2.672

1.693

[1, 6]

Leisure

2.10

3.27

[0, 55.7]

Durables

0.52

1.29

[0, 20.0]

Charity and gifts

3.40

14.47

[0, 501.4]

Personal care and clothing

2.70

3.44

[0, 82.8]

Health care

4.06

4.65

[0, 104.5]

Food in

3.92

4.49

[0, 97.2]

Food out

1.67

2.83

[0, 57.2]

Utilities and housing

17.88

19.14

[0, 352.5]

Vehicles

2.92

3.33

[0, 87.1]

Note: All expenditure measures, household income, household wealth are in thousands of 2006 dollars

3.3 Methods

This paper seeks to answer two research questions. The first is as follows: Which components of consumption expenditure are associated with increased subjective well-being?

To answer this research question, we estimate multivariate models, controlling for a wide array of household characteristics, relating happiness to consumption expenditure among older Americans using the HRS. In particular, we control for all categories of expenditure (in thousands of dollars) as well as age, income, wealth, marital status, race and ethnicity, gender and education, and a set of personality characteristics commonly referred to as the “Big 5” (neuroticism, agreeableness, extravertedness, conscientiousness, and openness).

In particular, we estimate the following linear regressions:
$$ {\text{Happiness}}_{i} = X_{i} \beta + \sum {\gamma_{j} \,{\text{Consumption}}_{ij} + \varepsilon_{j} } $$
(1)
where Happinessi is measured subjective well-being for individual i; Consumptionij is consumption expenditure on consumption category j by individual i; and Xi is a set of demographic, human capital, and personality variables for individual i.

Once again, the categories of consumption we examine include: leisure, durables, charity and gifts, personal care and clothing, health care, food in, food out, utilities and housing, and vehicles. These goods, in our view span the three possible channels through which consumption might affect happiness. Vehicle and personal care/clothing consumption are plausible “status” goods. Leisure consumption is a plausible “social” good. The remaining goods are likely related to happiness, if at all, through their effect on material well-being.

Identifying those categories of consumption that are related to happiness does not tell us why they are related to happiness or whether the consumption good in question is a “social” good, a “status” good, or neither. To partially address this issue, the second research question we ask is: To the extent that consumption expenditure is associated with happiness, is it the consumption of material goods itself or the social connections that come with certain forms of consumption that lead to happiness?

In particular, we add a set of psychosocial variables for depression, loneliness, and embeddedness in social networks to the models relating consumption expenditure and subjective well-being. To the extent that the effect of consumption expenditure on happiness is due to the social connections that are linked to consumption, we should observe these variables affecting happiness and the effect of consumption on happiness, in these models, to be diminished.

In particular, we estimate the following linear regression:
$$ {\text{Happiness}}_{i} = X_{i} \beta + \sum {\gamma_{j} \,{\text{Consumption}}_{ij} + \sum {\lambda_{k} \,{\text{SOCIAL}}_{ik} + \upsilon_{i} } } $$
(2)
where SOCIALik is a vector of social psychological variables (depression, loneliness, and embeddedness in social networks).

While we estimated Eqs. 1 and 2 using linear regression, as a sensitivity check (available upon request) we also estimated these models using an ordered probit model. All results were qualitatively similar whether estimated using an ordered probit or linear regression. Because of concern that collinearity among the consumption variables might contribute to large standard errors, we calculated the proportion of the total variation in each consumption variable that can be explained by the other explanatory variables, \( R_{j}^{2} \). These values range between 0.11 (for health care) to 0.58 (for leisure spending). Larger values of \( R_{j}^{2} \) can result in relatively high standard errors (Greene 2003).3

4 Results

Our findings suggest, first, that not all forms of consumption lead to happiness. In particular, only one category of consumption—leisure spending—has a statistically meaningful association with happiness (see Table 2, Model 1). In particular, a $10,000 increase in spending on leisure goods is associated with a 0.17-point increase in life satisfaction. None of the other components of consumption are significantly associated with life satisfaction with the exception of charity and gifts, which in some models has a very small but statistically significant association with happiness.
Table 2

Effects of consumption on happiness

 

Model 1

Model 2

Model 3

Model 4

Model 5

Leisure

0.170

(0.084)*

0.157

(0.083)

0.138

(0.083)

0.161

(0.084)

0.127

(0.083)

Durables

0.080

(0.199)

0.174

(0.189)

0.019

(0.197)

0.065

(0.203)

0.089

(0.194)

Charity and gifts

0.013

(0.008)

0.014

(0.008)

0.020

(0.008)*

0.014

(0.008)

0.020

(0.008)**

Personal care and clothing

0.126

(0.078)

0.124

(0.074)

0.138

(0.081)

0.122

(0.076)

0.131

(0.076)

Health care

−0.055

(0.078)

−0.043

(0.067)

−0.067

(0.072)

−0.062

(0.082)

−0.062

(0.067)

Food in

0.001

(0.072)

0.002

(0.067)

−0.013

(0.067)

0.006

(0.072)

−0.005

(0.064)

Food out

0.055

(0.156)

0.055

(0.172)

0.058

(0.156)

0.041

(0.158)

0.046

(0.169)

Utilities and Housing

−0.005

(0.013)

−0.007

(0.012)

−0.002

(0.012)

−0.003

(0.012)

−0.003

(0.011)

Vehicles

0.087

(0.052)

0.061

(0.048)

0.089

(0.055)

0.087

(0.051)

0.070

(0.049)

Depression (CESD)

 

−0.159

(0.017)**

  

−0.121

(0.017)**

Loneliness

  

−0.599

(0.057)**

 

−0.493

(0.057)**

Social interactions

   

0.046

(0.017)**

0.041

(0.016)*

Control for demographics and personality traits

Yes

Yes

Yes

Yes

Yes

R2

0.21

0.26

0.26

0.22

0.29

N

1,733

1,733

1,733

1,733

1,733

P < 0.05, ** P < 0.01, *** P < 0.001

Note: All expenditures are measured in tens of thousands of 2006 dollars. Control variables include age, income, wealth, marital status, race/ethnicity, gender and education. Personality traits include neuroticism, agreeableness, extravertedness, conscientiousness, and openness. Robust standard errors are reported in parentheses

This coefficient estimate suggests that a $10,000 increase in leisure expenditure is associated with roughly 14% of a standard deviation increase in happiness. While this association may appear modest, by comparison the association between being married and happiness is 0.346, suggesting that a $20,000 increase in leisure spending is roughly equal to the happiness boost one gets from being married.

Second, our findings suggest that leisure spending has a social or relational component. In Models 2 through 4, we add controls—one at a time—for depression, loneliness, and embeddedness in social networks. In Model 5, we add all three controls. Leisure spending is associated in part with higher levels of happiness through its social effect—by reducing loneliness and increasing embeddedness in social networks. This can be seen in Models 2 through 5.

Adding the control for depression reduces the coefficient on leisure spending very little (from 0.170 to 0.157), though it is no longer statistically significant at conventional levels in this model. Adding the control for loneliness reduces the association between leisure consumption and happiness lightly more, by about 19% (from 0.170 to 0.138), and it loses even more statistical significance. Adding the control for embeddedness in social networks reduces the association between leisure consumption and happiness by 5% (from 0.170 to 0.161). Adding all three reduces the association between leisure consumption and happiness by 25% (from 0.170 to 0.127). These results are all evidence of the “social” value of leisure consumption.

As discussed above, a large number of respondents have missing consumption data for some of the consumption categories. To determine the extent to which our results may be sensitive to this issue, we also used multiple imputation methods to impute missing consumption data and re-estimated our models using a sample that included imputed observations. The results for the association between leisure spending and life satisfaction are robust in these analyses. Similarly, no other category of consumption is associated with life satisfaction. However, in the multiple imputation models we find weaker evidence for the role of mediation by the psychosocial variables (results available upon request). In sum, we are confident about the association between leisure expenditure and happiness, but somewhat less confident about the role of the mediating factors.

5 Conclusion

We find that only one component of consumption expenditure is positively related to happiness—leisure consumption. Moreover the association between leisure spending and happiness is sizable: our estimates suggest that a $20,000 increase in leisure spending is roughly equal to the happiness boost one gets from being married. By contrast, all other components of consumption (which are primarily about meeting one’s material needs) appear to be unrelated to happiness. The finding that increased material well-being is unrelated to happiness lends support to the hypothesis (posed by Gilbert and Wilson 2007) that individuals quickly adapt in terms of their happiness to increases or to decreases in the level of their material well-being. Since our estimates are based on cross-sectional, not panel data, we are likely observing people at their steady-state level of happiness.

In our study, leisure consumption is associated with increased happiness to a moderate degree through its association with social connectedness, which we measure with three variables, depression, loneliness, and embeddedness in social networks. Interestingly, we found a role both for the perception of social connectedness (i.e., loneliness), and as for the actual quantity of time engaged in social activities and group functions. These findings provide new insights into how consumption may be associated with happiness.

However, the measures of social connectedness that we examined did not fully, or even mostly, explain why leisure spending is associated with happiness. This leads us to suggest that one can, to some extent, reject the conventional wisdom that “material goods can’t buy happiness,” insofar as many of the components of leisure we measure here are material goods (i.e., concert tickets, vacations, athletic equipment). Moreover, though many of the material goods in our leisure measure do have a relational component, it could also be argued that spending a lot on expensive vacations or donning the latest high-cost sporting equipment represents pure conspicuous consumption, designed to increase one’s status in the eyes of others.

Following this line of thought, if one considered the elements in our “leisure” category to reflect primarily “status” goods, then one could equally well conclude from our results that spending on status goods increases happiness in part through its effect on perceived connectedness. One might simply perceive a stronger sense of belonging or fitting in with the acquisition of status goods if, for example, such an acquisition makes one more similar to one’s reference group in terms of material purchases. In other words, “keeping up with the Joneses” might make a person feel more like “one of the Joneses” and hence less lonely and this might ultimately make him happier even if he never actually interacts with the Joneses. This idea corresponds to Layard’s (2005) contention that happiness is affected by an assessment of one’s own situation relative to one’s peers, but it also illuminates some of the psychological channels that might operate when individuals make such assessments.

The age structure of our sample may also be relevant to our findings. The average age of the respondents in these HRS data is 66 (with most of the sample ranging in age from 56 to 76). It could be that different kinds of expenditures bring happiness at different stages of life. For a relatively older adult, an expensive vacation or box seats at the symphony may be happily viewed as the just rewards of a long and productive life, something that was postponed until a stage of relative financial security and seniority in one’s position at work or in the community. Conversely, the quality of leisure experience could be more variable at older ages, thus dampening the correlation between leisure and happiness at this stage of life.

This research is subject to a number of limitations. First, the sample sizes on which these results are based are relatively small. Second, at this time data limitations prevent the use of panel data methods that would control for individual fixed effects. As more waves of the psychosocial questionnaire are made available, our sample size will increase and we will have multiple measures of the psychosocial variables across waves. Third, the fundamental issue of identifying causality is not addressed. We cannot determine whether leisure consumption is associated with happiness through its association with loneliness, for example, or whether loneliness is associated with happiness through its association with leisure expenditure.

Footnotes
1

In this paper we will also use these terms interchangeably.

 
2

In particular, we replace missing values for the consumption variables with the mean values for the sample. In all specifications, we also include a set of dummy variables indicating whether each consumption variable was imputed. In this way, we allow the relationship between happiness and consumption to differ between those with imputed and non-imputed consumption data.

 
3

The values of \( R_{j}^{2} \) for the consumption variables are: durables (0.13), utilities (0.14), health care (0.11), leisure (0.58), food at home (0.29), food away from home (0.33), charity and gifts (0.40), personal care (0.28), and vehicles (0.17).

 

Acknowledgments

The authors thank Milo Bianchi and participants at the 2009 Conference on Happiness and Relational Goods, held in Venice, Italy for helpful comments. We also thank Jen-Hao Chen for excellent research assistance.

Copyright information

© Springer-Verlag 2010