Subjective Well-Being: Keeping Up with the Perception of the Joneses

Abstract

Using data from the US General Social Survey 1972–2004, we study the role of perceptions and status in self-reported happiness. Reference group income negatively relates to own happiness and high perceptions about own relative income, quality of dwelling, and social class relate positively and very significantly to happiness. Perceptions about income and status matter more for females, and for low income, conservative, more social, and less trusting individuals. Dwelling perceptions matter more for males, and for middle income, married, conservative, more social, and less trusting individuals.

Appendix

1.1 Robustness Tables

We present the summary statistics for the variables used in the paper in Tables 11 and 12. We report the means for the continuous variables (for instance, weekly working hours) and proportions (for instance, labor force status) for the categorical variables.

Table 11 Other independent variables’ means, proportions, and standard deviations

Table 12 Other independent variables’ means, proportions, and standard deviations

We estimate self-reported happiness of an individual i at time t as a function of his/her ln perceived dwelling status with respect to other dwellings in his/her neighborhood and city, type of his/her dwelling, ln perceived social class, ln occupational prestige, ln perceived relative income, ln reference group income and ln household income together with the following controls variables: dwelling ownership, ln weekly working hours, labor force status, sex, age, age square, race, years of education, ln household size, marital status, religion, ln population of the place of residence, region fixed effects, year fixed effects, occupation fixed effects, and industry fixed effects.

Since the question on perceptions about relative income is asked with respect to “American families in general” in the GSS, we use ln GDP per capita as the reference group income in specification two. ln GDP per capita is negatively and significantly correlated with happiness, as expected, and it does not change any of our results. Then, in column 2, we use ln sector level (1-digit) wages and ln sectoral level GDP as reference group income. Perception variables do not change but total sector GDP does not predict happiness. In the fourth specification, we run an OLS regression of household income on parent’s education, spouse working hours, spouse labor force status and spouse occupational prestige. Then, we calculate the predicted income for everyone. We do the same in the fifth specification but now we use interval regression to calculate predicted income (original household income in the GSS is in intervals). Our results are robust to these cases but now reference group income positively correlates with happiness. Lastly, we use ln sectoral wage and predicted income together and again the results do not change. One concern in these regressions can be multi-collinearity. However, Table 14 shows that this is not the case and a perfect correlation does not exist among the independent variables used in the regression in Table 13.

Table 13 Alternative measures of household income and reference group income

Table 14 Correlation of instruments with main variables

We conduct robustness checks in several ways in Table 15. We control for ln personal income since perceptions might be correlated with personal income instead of household income. Our results again do not change. In the fourth specification, we estimate our model with ordered probit and find that the coefficients and the t-statistics are quite similar to the ones with OLS. We estimate our model for married and nonmarried samples and find that our results are again robust to different sub-samples. Next, we use levels of the income and status variables instead of logarithms in specification seven. In specifications eight and nine, we control for own and spouse’s socio-economic status which might affect our main variables of interest and happiness at the same time. The results again are robust.

Table 15 Robustness checks: I

We conduct more robustness checks in Table 16. We use perceived social class as a 10 category variable and then use happiness as a 7 category variable but our results do not change (these variables are often missing). In the third specification, we consider using “daily happiness” instead of general happiness. We find that daily happiness is not correlated with household income nor with actual relative income. However, perceptions about relative income and social class can predict daily happiness. Moreover, daily happiness is a 4 category variable which provides an opportunity to control for personal bias in choosing the middle category in odd numbered questions. In the fourth specification, we use residual happiness instead of actual happiness as the dependent variable. By doing so, we try to control for any collinearity problem between income and other variables since residual happiness is calculated after the regression of happiness on household income and other variables in the baseline regression. For example, if income aspirations are correlated with perceptions as well as happiness, this might bias our results. Therefore, we control for income aspirations. We also use the distance from poverty line as a measure of actual relative income and the results are robust to this (this variable is generated by the GSS). In addition, we control for spouse’s work status and use interval regressions to assign income levels from income intervals, assuming that income is log-normally distributed, rather than taking the midpoint in their conversion of categorical income categories to point estimates. The results are robust to these permutations.

Table 16 Robustness checks: II (dependent variable: self-reported happiness)

It is believed that childhood and past life events have effects on the current well-being of an individual. These events might also be correlated with perceptions which can bias our estimates. Therefore, we control for these variables in the regressions in Table 17. We find that individuals perceptions about their family income during childhood matters. It appears that the size of the coefficient on current perceived relative income declines with the inclusion of perceptions during childhood which suggests that there might be some persistence in perceptions during the life cycle (One might think of using perceptions during childhood as an instrument for current perceptions however the childhood perceptions are asked currently. Hence, current personal characteristics might influence the answer to this question. Therefore, we only consider using it as another explanatory variable for the current well-being.) Considering deaths, hospitalizations and divorces, we find that past events have strong influence on current happiness however we find evidence for adaptation. For instance, the influence of death of spouse in the last year is higher than the death fours ago or more than 6 years ago. Interestingly, people are affected more by the death of the mother than the father. There seems to be some adaptation to perceptions and also income shocks in the United States. We only display the coefficients on the extra controls in the table but our results are robust to the inclusion of past life events.

Table 17 Controlling for past life events and adaptation (dependent variable: self-reported happiness)

1.2 Variables Used in the Paper

Self-reported happiness: The answer to question 157 in the 2004 GSS codebook: “Taken all together, how would you say things are these days—would you say that you are very happy, pretty happy, or not too happy?” Answer: “not too happy” (1), “pretty happy” (2), “very happy” (3), and “don’t know” (missing), no answer (missing), not applicable (missing).

Perceived relative income: The answer to question 188 in the 2004 GSS codebook: “Compared with American families in general, would you say your family income is far below average, below average, average, above average, or far above average? Probe: Just your best guess.” Answer: “far below average” (1), “below average” (2), “average” (3), “above average” (4), “far above average” (5), and “don’t know” (missing), no answer (missing), not applicable (missing).

Perceived relative income when 16 years old: The answer to question 30 in the 2004 GSS codebook: “Thinking about the time when you were 16 years old, compared with American families in general then, would you say your family income is far below average, below average, average, above average, or far above average? Probe: Just your best guess.” Answer: “far below average” (1), “below average” (2), “average” (3), “above average” (4), “far above average” (5), and “don’t know” (missing), no answer (missing), not applicable (missing).

Perceived social class: The answer to question 185-A in the 2004 GSS codebook: “If you were asked to use one of four names for your social class, which would you say you belong in: the lower class, the working class, the middle class, or the upper class?” Answer: “lower class” (1), “working class,” (2) “middle class,” (3) “upper class,” (4) and “don’t know” (missing), no answer (missing), not applicable (missing).

Household income: 1) We created a continuous family income variable using the mid-point method to the the answer to question 37 in the 2004 GSS codebook: “In which of these groups did you total family income, from all sources, fall last year before taxes, that is? Just tell me the letter.” Answer: Under 1,000 (1), 1,000–2,999 (2), 3,000–3,999 (3), 4,000–4,999 (4), 5,000–5,999 (5), 6,000–6,999 (6), 7,000–7,999 (7), 8,000–8,999 (8), 10,000–14,999 (9), 15,000–19,999 (10), 20,000–24,999 (11), 25,000 and over (12), refused (missing), don’t know (missing), no answer (missing), not applicable (missing). 2) GSS generated, variable number 1437 in the 2004 GSS codebook, family income on 1972–2004 surveys in constant dollars (base 1986)

Health status: The answer to question 159 in the 2004 GSS codebook: “Would you say your own health, in general, is excellent, good, fair, or poor?” Answer: “Excellent” (4), “good” (3), “fair” (2), “poor” (1), and “don’t know” (missing), no answer (missing), not applicable (missing).

Marital status: The answer to question 4 in the 2004 GSS codebook: “Are you currently–married, widowed, divorced, separated, or have you never been married?” Answer: “married” (1), “widowed” (2), “divorced” (3), “separated” (4), “never married” (5), and no answer (missing). We recode this variable as follows: married = 1, and not married = 2, 3, 4, 5.

Labor force status: The answer to question 1 in the 2004 GSS codebook: “Last week were you working full-time, part-time, going to school, keeping house, or what?” Answer: “working full-time” (1), “working part-time” (2), “with a job, but not at work because of temporary illness, vacation, strike” (3), “unemployed, laid off, looking for work” (4), “retired” (5), “in school” (6), “keeping house” (7), “other” (8), and no answer (missing). We recode this variable as follows: employed = 1, 2, 3 unemployed = 4, and not in the labor force = 5, 6, 7, 8.

Working hours: The answer to question 1-A in the 2004 GSS codebook: “If working full or part-time: How many hours did you work last week, at all jobs? ” Answer is the number of hours, no answer (missing), not applicable (missing).

Sex: The answer to question 23 in the 2004 GSS codebook: Coded by the interviewer. “Male” (1) and “female” (2).

Race: The answer to question 24 in the 2004 GSS codebook: “What race do you consider yourself?” Answer: “white” (1), “black” (2), “other” (3), and not applicable (missing).

Education: The answer to question 15 in the 2004 GSS codebook: Coded as the number of years of schooling (maximum is 20) and don’t know (missing), no answer (missing).

Children: The answer to question 12 in the 2004 GSS codebook: ‘ ‘How many children have you ever had? Please count all that were born alive at any time (including any had from a previous marriage” Answer: 0, 1, 2, 3, 4, 5, 6, 7, 8 or more, don’t know (missing), no answer (missing).

Age: The answer to question 13 in the 2004 GSS codebook: Coded as the number of years from birth, don’t know (missing), no answer (missing).

Household size: The answer to question 34 in the 2004 GSS codebook: Coded as the number of household members (1–16), no answer (missing).

Region: The region of interview, question 51 in the 2004 GSS codebook: New England (1), Middle Atlantic (2), East North Central (3), West North Central (4), South Atlantic (5), East South Central (6), West South Central (7), Mountain (8), Pacific (9). (See Question 26 in the 2004 GSS codebook for a listing of the states within regions.)

Occupational prestige: The prestige scores assigned to occupations were taken from rating systems developed at (NORC) in a project on occupation prestige directed by Robert W. Hodge, Paul S. Siegel, and Peter H. Rossi. This concept of prestige is defined as the respondents’ estimation of the social standing of occupations. The prestige scores in the Hodge-Siegel-Rossi and GSS studies were generated by asking respondents to estimate the social standing of occupations in a nine-step ladder, printed on cardboard and presented to the respondent. It is constructed from two variables: (1) Occupational prestige score 1974–1992 on page 54 in the 2004 GSS codebook (2) Occupational prestige score 1992–2004 on page 57 in the 2004 GSS codebook. (See page 2050 for the construction of the prestige scores in the 2004 GSS Appendix.)

Occupation: One digit occupation categories for the respondents. It is constructed from two variables: (1) Occupational classifications 1974–1992 on page 54 in the 2004 GSS codebook (2) Occupational classifications 1992–2004 on page 56 in the 2004 GSS codebook. (See pages 2031 and 2068 for the subcategories of occupational categories in the 2004 GSS Appendix.)

Industry: One digit industry categories for the respondents. It is constructed from two variables: (1) Occupational classifications 1974–1992 on page 56 in the 2004 GSS codebook (2) Occupational classifications 1992–2004 on page 57 in the 2004 GSS codebook. (See pages 2055 and 2068 for the subcategories of occupational categories in the 2004 GSS Appendix.)

TV hours: The answer to question 242 in the 2004 GSS codebook: “On the average day, about how many hours do you personally watch television?” Answer: number of hours from 0 to 24, and don’t know (missing), no answer (missing), not applicable (missing). Reference for the variables and the explanations:

Political view: The answer to question 66-A in the 2004 GSS codebook: “We hear a lot of talk these days about liberals and conservatives. I am going to show you a seven-point scale on which the views that people might hold are arranged from extremely liberal–point 1– to extremely conservative–point 7. Where would you place yourself on this scale?” Answer: “Extremely liberal” (1), “liberal” (2), “slightly liberal” (3), “moderate, middle of the road” (4), “slightly conservative” (5), “conservative” (6), “extremely conservative” (7), and “don’t know” (missing), no answer (missing), not applicable (missing).

Socialization variables: The answers to question 173 in the 2004 GSS codebook: “Would you use this card and tell me which answer comes closest to how often you do the following things?” A)“Spend a social evening with relatives,” B) “Spend a social evening with someone who lives in your neighborhood,” C) “Spend a social evening with friends who live outside the neighborhood.” Answer: “Never” (1), “about once a year” (2), “several times a year” (3), “about once a month” (4), “several months a year” (5), “once or twice a week” (6), “almost every day” (7), and “don’t know” (missing), no answer (missing), not applicable (missing).

Dwelling variables: Dwelling type: The variable 1469 (interviewer coded) in the 2004 GSS codebook: Trailer (1), detached single family house (2), 2-family house, 2 units side by side (3), 2-family house, 2 units one above the other (4), detached 3–4 family house (5), row house (6), apartment house, 3 stories or less (7), apartment house, 4 stories or more (8), apartment in a partly commercial structure (9), other (10), don’t know (missing), no answer (missing), not applicable (missing). A): The answers to question 1470-A in the 2004 GSS codebook: “Compared to apartments/houses in the neighborhood, would you say the house/apartment was...” B) :The answers to question 1470-B in the 2004 GSS codebook: “Compared to apartments/houses in the city/town/county, would you say the house was...” Answers: “far below average” (1), “below average” (2), “average” (3), “above average” (4), “far above average”, and (5), “no answer” (missing), not applicable (missing). Dwelling own: The answers to question 1471 in the 2004 GSS codebook: “ (Do you/does your family) own your (home/apartment), pay rent, or what?” Answer: “own or is buying” (1), “pays rent” (2), “other” (3), and “don’t know” (missing), no answer (missing), not applicable (missing).

Mobility: The answers to question 26-A in the 2004 GSS codebook: “When you were 16 years old, were you living in this same (city/town/county)?” Answer: “same state, same city” (1), “same state, different city” (2), “different state” (3), and “don’t know” (missing), no answer (missing).

Religion: The answer to question 104 in the 2004 GSS codebook: “What is your religious preference? Is it Protestant, Catholic, Jewish, some other religion, or no religion?” Answer: “Protestant” (1), “Catholic” (2), “Jewish” (3), “none” (4), “other denominations” (5), “Buddhism” (6), “Hinduism” (7), “other Eastern” (8), “Moslem/Islam” (9), “Orthodox-Christian” (10), “Christian” (11), “Native American” (12), “Inter-Nondenominational” (13), and “don’t know” (missing), no answer (missing).

Socio-Economic Index scores: Socio-economic indexes of own (variable 1506 in the 2004 GSS codebook) and the spouse (variable 1510 in the 2004 GSS codebook) taking values 0–100.

Reference for the variables and the explanations:

Webpage: 1972–2004 GSS Codebook http://www.deakin.edu.au/~cahit/appendix.pdf

Webpage: 1972–2004 GSS Appendix http://www.deakin.edu.au/~cahit/codebook.pdf