Diagnosing our National Disease: Trends in Income and Happiness, 1973 to 2004
University of Pennsylvania
Direct correspondence to Department of Sociology, University of Pennsylvania, 3718 Locust Walk, Philadelphia, PA 19104; email@example.com.
An important fact of the happiness literature is the apparent disconnect between economic growth and happiness, referred to as the “Easterlin Paradox.” Although real income has grown over the last thirty years, happiness has stagnated or perhaps even declined. Drawing on the cumulative 1973 to 2004 General Social Survey and using a sample of working-aged adults, this article demonstrates the complexity of these trends and suggests that once we consider multiple sources of satisfaction, trends in real income have less paradoxical implications. The principal force behind declining happiness has been a decline in the number of working-aged Americans who are married, as well as declining marital satisfaction. These trends, however, have been largely independent of trends in income. Once marital factors are considered, the negative trend in happiness reverses direction, and economic factors emerge as the single most important force underlying growing happiness. Three other trends further suggest that income matters for well-being. First, trends in financial satisfaction have, in recent periods, overlapped with gains in real income. Second, perceptions of relative income have increased, despite growing income inequality. Third, there is no evidence for “overwork” among families, at least as applied to happiness.
The relationship between income and happiness invites a good deal of speculation. On the one hand, most evidence indicates that income and happiness are positively associated. This is perhaps unsurprising. Income is related to a variety of positive life outcomes, and, when describing the characteristics of a good life, most people identify income as a key component (Diener and Oishi 2000). Yet two other features of the relationship remain puzzling. First, the relationship between income and happiness is characterized by diminishing returns. For example, those appearing on the Forbes magazine’s list of wealthiest Americans report only slightly more happiness than the average American, and some, in fact, report being quite unhappy (Diener, Horwitz, and Emmons 1985). Although this pattern is consistent with theories emphasizing the declining marginal utility of income (Veenhoven 1991), it is paradoxical given the many opportunities wealth would ordinarily seem to provide. Second, and related to the first, the positive relationship between income and happiness found in the cross-section does not appear to hold over time. Real income has increased in the United States over the last thirty years, and, in a very real sense, the material conditions of life have improved (Templeton 1999). Nevertheless, happiness has, if anything, declined, a phenomenon now referred to as the “Easterlin paradox” (Blanchflower and Oswald 2004; Diener and Suh 1997; Easterlin 1974, 1995).
These two paradoxes of income have inspired wide-ranging speculation. Some emphasize psychological factors, such as the persistence of relative deprivation. Others emphasize rising consumption needs, in particular the hunger for luxury items that may bring status but perhaps do little in the long term for well-being. Still others emphasize the behaviors associated with economic growth, arguing that the income growth has come at the expense of leisure, relationships, and health. Many of these arguments, at least in their general form, are long-standing. More than a century ago in a letter to H.G. Wells, William James bemoaned America’s “exclusive worship” of monetary success, which he identified, with great disgust, as our “national disease” (from letters collected in Skrupskelis and Berkeley 2003).
But is success, in fact, our national disease? Has our pursuit of wealth, in effect, derailed our pursuit of happiness? Although speculation is common, empirical research, especially research on trends, is surprisingly rare. In this article, I use the cumulative 1973 to 2004 General Social Survey (GSS) to address trends in happiness. Because the GSS is a repeated cross-section gathered over many years, it allows me to explore trends in an especially fine-grained fashion. Furthermore, the GSS provides a variety of relevant explanatory variables, including work hours and assorted financial evaluations. Indeed, this set of variables is exhaustive in a critical sense—with these variables, I am able to explain the entire income-happiness relationship and, at the same time, explain why happiness has not improved over time.
THEORIES LINKING INCOME AND HAPPINESS AND EXPLAINING WHY HAPPINESS HAS NOT IMPROVED
The relationship between income and happiness is usually explained in terms of income, allowing the individual to pursue courses of action and consume goods that improve well-being. As most scholars now recognize, however, this intuition obscures the actual complexity of the relationship. Rather than a more-or-less direct pathway from income to consumption to happiness, evidence reveals assorted indirect pathways, with income passing through judgments regarding one’s finances relative to one’s needs, as well as the behaviors that lead to more income but may or may not lead to happiness (Johnson and Krueger 2006). For those interested in the Easterlin paradox, the problem lies in distinguishing these often-conflicting trends.
The effects of income may depend on judgments regarding whether that income is adequate for maintaining a desired standard of living (Campbell, Converse, and Rodgers 1976). Judgment models refer to a general class of psychological processes wherein the individual evaluates an object relative to a standard. In the case of income, the relative-deprivation approach is perhaps the most common such model and one of the most popular explanations for the Easterlin paradox (see Festinger 1954 for an early statement). In this approach, the relevance of income for happiness rests with the status it confers rather than the consumption it allows. While a given level of income may be adequate from the standpoint of meeting all basic needs, it might nonetheless fail to improve well-being if that income is perceived to be below some status-centered norm. Because happiness is positively related to personal income but inversely related to the income of others, perceptions of relative standing tend to remain constant over time and so deflate the gains in utility one might otherwise realize from economic growth.
Other judgment models focus less on status and more on satisfaction. The effects of financial satisfaction in mediating the income-happiness relationship are, in most studies, quite large (George 1992). In the income literature, the prevailing view suggests an aspiration treadmill, wherein aspirations change as fast as actual circumstance, leading to a weak association between financial satisfaction and income—individuals always have slightly less than they want. (van Praag and Frijters 2003). Consistent with this view, there is a good deal of evidence that material aspirations shift commensurate with wealth. For example, public-opinion surveys regarding the elements of a “good life” reveal that the number of goods one desires grows in a parallel fashion with the number of goods one already owns (Easterlin 2005). Similarly, the income individuals report as necessary to “get along” in life is hardly stable and, above all, increases with income (Rainwater 1974). Others have made very similar arguments, noting that our aspirations may have, in fact, increased much faster than our income. Along these lines, Frank (1999) argues for an increasingly status-focused culture, wherein the pursuit of luxury goods is increasingly important to self-respect. In a similar vein, Schor (1999) argues that Americans are increasingly concerned with emulating the lives of the rich and famous. Of course, these things not only involve a change in how we evaluate our own status; they may also entail changes in our behavior.
Overwork and the Systematic Underinvestment in Leisure
Broadening the scope of economic approaches to happiness, a different approach to the Easterlin paradox emphasizes the trade-offs individuals make in pursuit of higher income. This approach is premised on two ideas: one, that individuals hold various misconceptions regarding the actual determinants of happiness and, two, that when pursuing wealth, individuals make decisions that systematically yield unhappy results (see Gilbert 2006 for a summary of affective forecasting). Although not supported by any direct evidence, the idea that Americans are increasingly allocating their time in an ineffective manner is supported by assorted pieces of indirect evidence. At a basic level, we know that Americans overvalue income relative to other demonstrable contributors to happiness. In studies asking respondents to list the key ingredients to happiness, income consistently ranks high, often higher than family or health (Cantril 1965, but see Kasser and Ryan 1993 on finances vs. self-acceptance among college students). In the same vein, there is some evidence linking materialism to impaired relationships. Sheldon and Kasser (1995) find that those oriented toward extrinsic goals, including wealth accumulation, tend to have more conflict-ridden relationships with family and friends.
Perhaps even more compelling, at least for the present study, survey data reveal the kinds of trends and value shifts that might deflate happiness over time. From the early 1970s to the mid-1980s, the stated importance of private materialism increased sharply, while the importance of self-fulfillment—indicated by support for things such as finding “purpose and meaning” in life—declined (Easterlin and Crimmins 1991). Likewise, McPherson, Smith-Lovin, and Brashears (2006) find an increase in social isolation between 1985 and 2004, and speculate, as have others (Putnam 2000), that this increase may be due to growth in work hours among families. This confluence of factors may be sufficient to produce behavior that appears irrational with respect to the actual determinants of well-being. However, the idea that Americans allocate time in suboptimal ways has not been fully evaluated. In fact, we know little about the relationship between work hours and happiness, let alone the relationship between work hours and various other contributors to happiness.
This lack of empirical analysis represents a more general problem in the income-happiness literature. The field has a number of well-established theories for explaining why happiness has not increased even as real income has grown and each has received at least some empirical support. Yet few scholars have tested one theory against the other. In this light, it remains unclear which theory is most important for understanding trends and we are left with a cluttered field of contending theories. My empirical analysis is devoted to understanding why happiness has or has not changed over time in response to changes in income. In answering this question, I first address the empirical viability of each of the three theories discussed above and then apply these theories to understanding trends in happiness.
DATA AND METHODS
I use the cumulative 1973 to 2004 General Social Survey (GSS) (Davis, Smith, and Marsden 2004). The GSS is one of the most well-regarded and influential data sources in the United States. The survey is conducted face-to-face and is representative of the non-institutionalized adult population of the United States. Because the GSS has been fielded on a regular basis since the early 1970s, the data are especially useful for evaluating trends. Although a longer period of time would have allowed for a more fine-grained investigation—capitalizing, for example, on long-term swings in economic performance—the period covered here has been the focus of much of the debate.
Appendix Table A (appearing at the end of this article) describes the key variables. The table presents the years and samples in which each of the items appeared, as well as a basic description of the survey questions. The present study is concerned with evaluating trends, so the bulk of the analysis focuses on questions asked repeatedly and consistently. These include: a question about overall happiness; measures of family income, adjusted for inflation; satisfaction with one’s financial situation; perceived relative income; marital happiness; self-rated health; and average family work hours per week. The models include other controls, but these serve as the central elements of my analysis.
The analysis begins with two multiple-panel figures summarizing basic patterns. Figure 1 displays four patterns: the relationship between real family income and happiness, trends in income, trends in happiness, and trends in total family work hours. Among other things, Figure 1 reveals a reversal in the happiness trend, the first sign—of several more to come—of an alignment between trends in income and trends in well-being. Two things are important to appreciate, as they are consistent with previous research. First, the relationship between income and happiness, while positive, diminishes at higher levels. Second, trends in happiness do not reflect trends in real income, at least not in any obvious way. Despite gains in real income throughout the 1980s, happiness declined from 1973 through the mid-1990s. After that, happiness began, somewhat suddenly, to increase.1 This result is consistent with both the Easterlin paradox and research showing a slight positive increase in happiness in developed countries. Given the positive relationship between income and happiness, this late-period increase might very well reflect real economic growth. Adding to this possibility, total family work hours declined somewhat after the mid-1990s, perhaps suggesting a softening of any trade-offs between work hours and well-being. Nevertheless, trends in happiness prior to 1994 are more complex and clearly cannot be explained by trends in real income alone.
FIGURE 1. Key Patterns in Happiness and Real Family Income, 1973 to 2004 General Social Survey
1 To avoid, for the moment, confounding trends in real income with other demographic trends, the panels displaying trends in real family income and trends in total family work hours are restricted to married respondents. The regression models, however, use the full sample.
Figure 2 turns to the key mediating variables for the income-happiness relationship. The figure presents the relationship between real income and each of the mediating variables. Recall that the theories outlined above suggest a weak relationship between income and several of the mediating variables or a relationship that diminishes with increasing income. For example, income should not have an especially strong association with financial satisfaction, given adaptation. In Figure 2, three things are important to appreciate. First, there is no mediating variable for which the relationship reaches a true plateau. The highest expected value is always found at the highest level of income. Second, not every relationship is characterized by equally steep diminishing returns. The slope between income and financial satisfaction, for example, diminishes very slowly, as does the slope between income and perceived relative status. The former is not entirely consistent with the idea of adaptation and allows for the possibility that financial satisfaction has increased with improvements in real income. Third, all the relationships are not equally strong, which suggests that understanding happiness, including its trends, requires more than understanding income and its correlates. In fact, the relationship between income and marital happiness is very weak (.076), whereas the relationship between income and financial satisfaction is quite strong and shows little in the way of diminishing returns.
FIGURE 2. Relationships between Real Family Income and Key Mediating Variables, 1973 to 2004 General Social Survey
Table 1 examines trends in the key mediating variables: perceived relative income, financial satisfaction, and, because of their relationship with work hours, marital happiness and health. In all the models, year is modeled using a two-covariate spline function with knots set to overlap with the trends revealed in Figure 1. The coefficients in the upper panel are from a multinomial logit model predicting perceived relative income. The four equations predict the likelihood of reporting the given category of relative income relative to reporting “average” income. The model controls for actual family income. As a result, the year coefficients can be interpreted as the per-decade change in the likelihood of reporting a given level of relative income, holding actual income constant. For the same reason, I also control for income when examining financial satisfaction.
Above all, Table 1 reveals countervailing forces, many of which are inconsistent with expectations derived from the theories described above. For example, between 1973 and 1994, Americans grew increasingly likely to report above average income. Yet, at the same time, Americans grew increasingly dissatisfied with their finances, and both marriage rates and marital happiness declined (marital status and happiness are here disentangled in order to clearly demonstrate trends). Furthermore, some of these trends shifted after 1994, consistent with trends in happiness. After 1994, marital happiness began to increase, as did financial satisfaction (although not significant in the table, the 1994 to 2004 coefficient is significant if real income is not included in the model). Which of these trends is most important for understanding why happiness has not increased?
Table 2 attempts to explain trends in happiness by using multiple regression. Table 2 presents five models with progressively more stringent controls. Together these models suggest that the apparent stability of happiness amidst rising income is an artifact of countervailing trends. Model 1 begins with two key relationships: the decline in happiness found between 1973 and 1994 and the gradually decelerating relationship between income and happiness. These are modeled using two coefficients, the first for the trend to 1994, which is negative, and the second for the trend thereafter, which is positive. The remaining models attempt to explain these twin patterns. A satisfying model would reduce either or both of these coefficients to zero and/or statistical insignificance. Model 1 controls for income, as well as age, race, and sex. Income is modeled as three separate spline components, corresponding to tertiales of low, medium, and high. As expected, the coefficient for the lowest coefficient is largest, suggesting that the income-happiness relationship is stronger at lower levels of income, meaning that happiness necessitates a basic minimum of income. Model 2 is the first attempt to explain the trends and, to that end, adds perceived relative income. Perceived relative income does little to explain trends in happiness—as noted above, perceived relative standing has, if anything, increased over time—but it does explain a large fraction of the income-happiness relationship, especially at lower levels of income.2 In Model 3, I add financial satisfaction. Financial satisfaction explains two-thirds of the remaining income-happiness relationship in the third tertiale. Financial satisfaction also explains about 34% of the decline in happiness between 1973 and 1994: the coefficient drops from -.029 to -.019. Nevertheless, the negative trend in happiness remains significant.
2 In addition, because the two coefficients for below-average income are much more negative than the two coefficients for above-average income are positive, the mediating effects of perceived relative income are greater at lower levels of income. For example, with controls for relative income, the coefficient for the first tertiale of income is reduced by 34%, whereas the coefficient for the third tertiale is reduced by less than 20%.
Model 4 presents a very different picture. Controlling for marital happiness (which has marital status embedded within it) reverses the trend in overall happiness: the trend coefficient up to year 1994 shifts from -.019 to .027. This can be interpreted in a counterfactual fashion: if it was not for declines in marital happiness between 1973 and 1994—that is, if everyone were assigned the average—overall happiness would have increased by about the same amount as it actually appeared to decline. Part of the reason for this reversal is the magnitude of the marital happiness-overall happiness relationship. The coefficients for martial happiness are remarkably large, suggesting that marriage is very important for well-being. To be sure, only those reporting “very” happy marriages are happier than those who are not married, but many Americans do in fact, report very happy marriages, so the average marriage is strongly associated with happiness. Model 5 adds controls for self-rated health and, thus, explores the role of potential trade-offs between work hours—which have propelled gains in real income—and health. Perhaps Americans are overworking and, so, increasing their income at the expense of their health. With this model, I am able to explain the entire remaining income-happiness relationship.
Table 2 reveals a number of things, but most importantly it reveals that trends in marriage and marital happiness are offsetting otherwise positive trends in happiness. Other models further support the idea that Americans would be happier, were it not for these trends in marriage. I explored the relationship between family work hours and happiness, for example, as well as the relationship between family work hours and other sources of satisfaction, including marital happiness, satisfaction with friendships, and self-rated health. The results generally support the idea that long work hours compromise well-being, as expected, but they also suggest that the point at which work hours begin to compromise health and well-being is well in excess of 80 hours per week. Recall from Figure 1 that the average work week per family exceeded 65 hours only once during the entire period under consideration. If anything, Americans are moving closer to the optimum, not further from it. Furthermore, if one begins with models that control for marital status and happiness and, thus, show a positive increase in happiness over time, income and employment play a much more decisive—and less paradoxical—role. Controlling for income, employment status, and work hours explains much of the increase in happiness found between 1973 and 1994.
The results of this study point to a rather unexpected alignment between economic behavior and happiness. Happiness has, indeed, declined, but this decline appears to have less to do with financial judgments or overwork than trends in marital status and marital satisfaction. Furthermore, the paradox of rising income and declining happiness does not appear to hold once marital factors are considered. In fact, after including these factors, overall happiness increased between 1973 and 2004, and this increase can be understood in a rather straightforward fashion, with little recourse to psychological judgments or status. Virtually all of the increase is attributable simply to rising income and growing labor-force participation. Furthermore, there is no evidence that individuals are working in ways that compromise their well-being. Indeed, was it not for growing economic opportunity, the twenty-year decline in happiness found between 1973 and 1994 would have been a good deal more severe. By the same token, the especially strong increase in happiness beginning in the late 1990s appears to be due to gains in real income, complemented by growing financial satisfaction. During this period, real income and financial satisfaction moved in tandem, despite speculation regarding America’s growing need for status and luxury goods.
All this is not to say that any one of the three frameworks outlined in the introduction is entirely incorrect—it is merely to suggest that the direct pathway from income and happiness deserves more credit than it is usually given. Other explanations should be put in their broader context. Relative deprivation, for example, has a powerful relationship with well-being: net of actual income, those who believe their income is below average are much less happy than those who believe their income is average. Yet from the standpoint of declining happiness, relative deprivation is rather unimportant. In fact, Americans grew increasingly likely to report above average income between 1973 and 1994, even as their happiness fell somewhat and wealth became, at least according to some, more ostentatious. This result begs a number of questions, perhaps the most important of which is the source from which Americans derive their standards of comparison. It is entirely possible that Americans derive their standards from personal past income as much as the putative current income of others. Similarly, financial satisfaction is strongly related to happiness and, as much of the literature anticipates, it does not track perfectly with trends in actual income. Between 1973 and 1994, for example, financial satisfaction declined at approximately the same pace as marital happiness declined. Satisfaction, in short, comes from something other than income alone. Yet even with this growing financial discontent, financial satisfaction explains no more than 35% of the decline in overall happiness. By the same token, scholars are correct to assert that there are trade-offs between work hours and well-being, but those families working long enough to compromise their well-being are clearly outliers. The average American family is very different. Indeed, trends in labor-force participation and work hours have if anything propelled, not undermined, happiness.
Above all, this study reveals that the “paradox” of income may have been overdrawn. In fact, over the last thirty years, happiness would have declined a great deal more than it did were it not for gains in real income. This is not to say that all Americans are, under all circumstances, acting in ways that maximize their well-being. Nor is it to suggest that the pursuit of wealth never entails a shadow side—from an empirical standpoint, commentators like William James are correct to highlight the perils of “worshiping” financial success, especially exclusively. But it is to suggest that Americans may derive more satisfaction from income than is often acknowledged and, moreover, they may do a good job, on average, of balancing a happy life.
Blanchflower, David G. and Andrew J. Oswald. 2004. “Well-Being Over Time in Britain and the USA.” Journal of Public Economics 88:1359-86.
Campbell, Angus, Philip E. Converse, and Willard L. Rodgers. 1976. The Quality of American Life. New York: Russell Sage.
Cantril, Hadley. 1965. The Pattern of Human Concerns. New Brunswick, NJ: Rutgers University Press.
Davis, James A., Tom W. Smith, and Peter V. Marsden. 2004. General Social Surveys, 1972 to 2004. [CUMULATIVE FILE] [Computer file]. Chicago, IL: National Opinion Research Center [producer]. Storrs, CT: Roper Center for Public Opinion Research, University of Connecticut/Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributors].
Diener, Ed, Jeff Horwitz, and Robert A. Emmons. 1985. "Happiness of the Very Wealthy." Social Indicators Research 16:263-74.
Diener, Ed and Shigehiro Oishi. 2000. "Money and Happiness: Income and Subjective Well-Being Across Nations." Pp. 185-218 in Culture and Subjective Well-Being, edited by E. Diener and D. M. Suh. Cambridge, MA: MIT Press.
Diener, Ed and Eunkook M. Suh. 1997. "Measuring Quality of Life: Economics, Social, and Subjective Indicators." Social Indicators Research 40:189-216.
Easterlin, Richard A. 1974. "Does Economic Growth Improve the Human Lot?: Some Empirical Evidence." Pp. 89-125 in Nations and Households in Economic Growth, edited by Paul A. David and Melvin W. Reder. New York: Academic Press.
———. 1995. "Will Raising the Incomes of All Increase the Happiness of All?" Journal of Economic Behavior and Organization 27:35-47.
———. 2005. "Building a Better Theory of Well-Being." Pp. 29-64 in Economics and Happiness: Framing the Analysis, edited by Luigino Bruni and Pier L. Porta. New York: Oxford University.
Easterlin, Richard A. and Eileen M. Crimmins. 1991. "Private Materialism, Personal Self-Fulfillment, Family Life, and Public Interest: The Nature, Effects, and Causes of Recent Changes in the Values of American Youth." Public Opinion Quarterly 55:499-533.
Festinger, Leo. 1954. “A Theory of Social Comparison Processes.” Human Relations 7:117-40.
Frank, Robert H. 1999. Luxury Fever: Why Money Fails to Satisfy in an Era of Excess. New York: Free Press.
George, Linda K. 1992. "Economic Status and Subjective Well-Being: A Review of the Literature and an Agenda for Future Research." Aging, Money, and Life Satisfaction: Aspects of Financial Gerontology, edited by Neal E. Cutler, Davis W. Gregg, and M. Powell Lawton. New York: Springer.
Gilbert, Daniel T. 2006. Stumbling on Happiness. New York: Knopf.
Johnson, Wendy and Robert F. Krueger. 2006. "How Money Buys Happiness: Genetic and Environmental Processes Linking Finances and Life Satisfaction." Journal of Personality and Social Psychology 90:680-91.
Kasser, Tim and Richard M. Ryan. 1993. “A Dark Side of the American Dream: Correlates of Financial Success as a Central Life Aspiration.” Journal of Personality and Social Psychology 65:410-22.
McPherson, Miller, Lynn Smith-Lovin, and Matthew B. Brashears. 2006. "Social Isolation in America: Changes in Core Discussion Networks Over Two Decades." American Sociological Review 71:353-75.
Putnam, Robert D. 2000. Bowling Alone. New York: Simon and Schuster.
Rainwater, Lee. 1974. What Money Buys. New York: Basic Books.
Schor, Juliet B. 1999. The Overspent American: Why We Want What We Don't Need. New York: Harper Perennial.
Sheldon, Kennon M. and Tim Kasser. 1995. "Coherence and Congruence: Two Aspects of Personality Integration." Journal of Personality and Social Psychology 68:531-43.
Skrupskelis, Ignas K. and Elizabeth M. Berkeley. 2003. The Correspondence of William James. Charlottesville, VA: University of Virginia Press.
Templeton, John M. 1999. "A Worldwide Rise in Living Standards." The Futurist 33:17-22.
van Praag, Bernard M. S. and Paul Frijters. 2003. "The Measurement of Welfare and Well-Being: The Leyden Approach." Pp. 413-33 in Well-Being: The Foundations of Hedonic Psychology, edited by Daniel Kahneman, Ed Diener, and Norbert Schwarz. New York: Russell Sage.
Veenhoven, Ruut. 1991. "Is Happiness Relative?" Social Indicators Research 24:1-34.