Religiosity, Economics and Life Satisfaction

Transcription

1 Religiosity, Economics and Life Satisfaction Kenneth V. Greene Bong Joon Yoon Department of Economics State University of New York at Binghamton Binghamton, New York April 2003

2 I. Introduction 1 Economists recently have begun to turn back to their roots in two areas: the study of the economic dimensions of religion and the exploration of human happiness. Their major "literature review" journal has recently published important pieces on each area [Iannaccone (1998), and Frey and Stutzer (2002)]. Our purpose is to provide a contribution to the literature which interfaces between religion and happiness. Economists have tended to focus on whether income or wealth produce greater happiness and their literature on religion while diffuse has tended to stress the potential material advantage of religious practice in reducing delinquent and criminal activity [Freeman (1986), Lipford, McCormick and Tollison (1993), Hull and Boyd (1995), and Hull (2000)]. It was not always so. Adam Smith noted that religion prepares people "for another and better world to come" [Smith ([1776]1981, p.778)] and that its current life benefits are often clearly psychological and happiness inducing. So he wrote about the rural migrants arrival in the city in the beginning of the industrial revolution. While such a migrant is initially "sunk in obscurity and darkness", often eventually "his conduct never excites so much the attention of any respectable society, as by becoming the member of a small religious sect" wherein he "acquires a degree of consideration which he never had before". [Smith ([1776]1981, p.795)]. Non-economists, however, have focused on a connection between religion and happiness. A recent survey is contained in Argyle (1999) and fairly recently Ellison (1991) has focused on the role of religious beliefs in happiness using American data. Our purpose is to supplement his contribution by focusing on European data and at the same time controlling for a variety of phenomena that economists have considered important determinants of happiness. At the same time, we want to question some conclusions that economists have reached while neglecting the role of religion. We do so by providing a brief review of the relevant literature in the next section. Section three spells out our model. It is followed by a description of our data. Section five presents and discusses our results. A concluding section follows.

3 II. Review of the Contributions of Economists to the Study of Happiness and the Role of Religion 2 Since Frey and Stutzer (2002a; 2002b) and Alesina, DiTella and MacCulloch (2001) spend considerable effort justifying subjective happiness research and showing that objective extra-personal measures of well-being correlated strongly with subjective happiness and much justification is provided in Kahneman, Diener, and Schwartz (1999), we will not do so. Nor shall we spend time on the voluminous literature on the "contradiction" that income increases happiness cross sectionally but perhaps not for the same individual across time [see Hollander (2001)]. We merely mention several consistencies that Frey & Stutzer emphasize and several that they do not. The first is that at any point in time people with more income declare themselves happier or more satisfied. Another is that a person's own unemployment decreases happiness. A third is that political, economic and personal freedom apparently imply greater individual happiness 1. Finally, they stress that macroeconomic conditions like inflation and overall unemployment may also affect individual satisfaction. Some consistencies they do not emphasize are that femaleness, age 2, marital status and education are positive determinants of subjective well-being. One of the important studies by economists (Alesina et.al 2001) also tests the hypothesis that income inequality is a public bad and reduces individual welfare as measured by subjective happiness. While they find no evidence for this in American data they do find that in Europe those with left wing political identification are negatively influenced by income inequality 3. One major study linking religion and subjectively assessed well being is cited by Iannaccone and appears in a social behavior journal. Ellison (1991) studies a small group of individuals in the United States who in the 1988 General Social Survey answered questions not only about the usual socioeconomic characteristics and membership in voluntary associations, and religious affiliation, but also question about recent traumatic events 4, religious attendance, feelings of closeness to a divine entity (divine interaction) 5 and certainty of belief (existential certainty) 6. Holding age, sex, income, race,

4 3 marital status, urban residence and region constant, Ellison uses ordinary least squares to explain life satisfaction 7. The usual significant results apply but he also finds that some religious affiliations increase life satisfaction 8 and that belief in divine interaction increases satisfaction and that in times of trauma existential certainty also does so. Thus the evidence points to a significant positive effect of what might be called "faith" or "religiosity". III. The Model The economics literature has assumed that subjective well being is a function of certain personal characteristics, certain narrowly economic characteristics of the individual, certain peculiarities of time and place, some macroeconomic characteristics of the society lived in and in some cases, certain political and societal characteristics that increase individual autonomy. More formally: [1] SWB i = SWB i (P i, E i, T i, M i, PL i G i ) where SWB i is expressed subjective well being of individual i, P i is a vector of her personal characteristics, E i are her economic characteristics, T i is the survey year, M i are macroeconomic conditions. PL i is the place of residence and G i are the governmental characteristics of the polity and society to which the individual belongs. The literature on religion and personal welfare has assumed that there is one more term in the equation explaining subjective well being - the religious characteristics of the individual R i - but also two less -- the macroeconomic and political conditions or: [2] SWB i = SWB i (P i, E i, T i, PL i, R i )

5 4 What we intend to explore is the possibility that expressed subjective well being is a function of all seven elements or: [3] SWB i = SWB i (P i, E i, T i, M i, PL i, G i, R i ). Our statistical model assumes a random error term, e i with mean zero. It also recognizes the idea that subjective well-being or satisfaction is recorded as a discrete variable taking on various values. In our case they are between 1 and 4 and as a result the proper statistical formulation for testing the model via regression techniques is an ordered logit specification. 9 IV. Data and Statistical Procedure Our survey data is from the Euro Barometer Survey (ICPSR, 1993) and contains answers for adult individuals in thirteen Western European countries from 1975 through The main question that we focus upon is a current life satisfaction response. That data set also includes most of the variables we use. They are defined in Table 1. In order to test some of the hypotheses developed in the economics literature, we also call upon data on the rate of inflation, the real gross domestic product, and the unemployment rate in the country by year. These have been garnered from the OECD and The World Output Tables. Again to address some questions which have interested economists, we use data on the Gini ratios for income inequality presented by Deininger and Squires (1996). Since the life satisfaction variable can take on one of 4 discrete values, we use the ordered logistic specification utilized by Alesina et al. To begin with, we simply repeat their experiment on the entire sample including as determinants of life satisfaction a person's employment status, sex, age, education, marital status, family size, income quartile, year and country of residence. The sample size is 274, 388.

6 5 We consider the influence of religious phenomena per se on self-perceived satisfaction and, as best we can, control for the possibility that some of the religious variables may pick up the effect of social participation and living in smaller places. Moreover we wish to add the macroeconomic conditions that economists have reasoned could affect satisfaction while controlling for religion. This results in a sample of 27,100. We also want to test the hypothesis developed by Buttle, Wilkening, and Martinson (1977) that the individual's political stance might affect perceived satisfaction and the hypotheses of Alesina et al. (2001) that macroeconomics effects depend on political stance, so we run the same specification on a sample of 27,100 but control for left and right wing political associations. 10 Finally, we attempt to see whether the results are similar for those that are left wing, right wing or neither. These samples contain 7,762, 7,850 and 11,488 (7, , ,488 = 27,100) observations respectively. V. Discussion of the Results Table 2's column 1 displays our results on the entire sample. Country and year dummies, though mostly all significant, are suppressed for clarity of presentation. The person's employment status is a significant positive determinant of perceived satisfaction. Females are significantly more satisfied as are those married and those more educated. A higher place in the income distribution increases satisfaction. It decreases from age 19 through roughly age 42 and then increases. Having more children significantly decreases happiness. So all told the economic variables work as expected. In column 2, we turn to the sub-sample for which we have observations on religious phenomena, macroeconomic characteristics, political status, and our proxies for social participation. The labor force

7 6 status (except for unemployment) and the number of children below 3 variables are no longer significant. Still, there is relatively little difference in terms of statistical significance for the other original variables. The religious variables included are nominal Catholicism, Protestantism, Orthodoxism or other religion. None of these has any significant effect. We also have two other variables for each religion that are indicators of more commitment. The first is attending services at least once a week and the second is attendance even more frequently. All such attendance variables do have a significant positive effect on perceived satisfaction. The coefficients for the second are always greater and, moreover, for Protestants the coefficient for the greater commitment is significantly more positive than that for weekly attendance. Apparently real, effective religious commitment increases life satisfaction even in Europe. This corroborates the results in the United States. Of course, the results of Ellison for the U. S. are stronger evidence for the effect of religion because he can control not only for attendance but also for elements of "faith". Moreover, he controls for the number of memberships the individual has in voluntary associations. 11 Our only control is whether or not the individual lives in a big place and whether he often engages in political discussions. Though the former is significantly negative 12, the latter, while positive, lacks significance. In Table 2, the results for the sample of 27,100 control for the political affiliation. Others [Buttel et al. (1977)] have argued, left-wingers tend to be less satisfied and right-wingers more so. Our results corroborate this, though the right variable is not quite statistically significant at the conventional significance level. Note also that column 2 in Table 2 reveals a negative effect of income inequality in the sample where we can control both for macro conditions and religious phenomena. This corroborates the finding of Alesia that inequality is perceived to reduce perceived life satisfaction in Europe. We also attempted to see whether religious attendance had different effects for different political affiliations. As one can see from Table 3, while very frequent church attendance appears to increase the perceived satisfaction of Catholic leftists more than Catholic neutrals and rightists, the opposite is true for

8 7 very frequent attendance by Protestants. It is the right-wingers whose satisfaction is more enhanced. But none of these differences are statistically significant. We also wish to return to an analysis of the effects of macroeconomic variables on life satisfaction emphasized by Alesina et al. (2001) and utilized by DiTella, MacCulloch and Oswald (2001) to reach the conclusion that economic misery indices should not simply add inflation and unemployment rates but must weigh the latter more heavily. In neither Table 2 nor 3 do we see the greater statistical significance of unemployment versus inflation in reducing life satisfaction. Table 3 implies that all political persuasions are negatively affected by inequality, though the effect is stronger for those on the left. Apparently, recognizing the important role of religion upsets the conclusions that overall unemployment must be weighted more than inflation and only leftists feel unequal societies reduce satisfaction 13. Table 3 which looks at people with different political orientations separately maintains these conclusions about the lack of significance of the macroeconomic variable. The sole exception is that right-wingers become less satisfied if inflation rises. Again, returning to the role of religion, we earlier discussed the relative strength of Ellison's American results. Let us note here that in some senses our findings are stronger for the thesis that religious experiences are important for life satisfaction. Our European sample allows us to correct for many more potentially important economic phenomena and allows a vastly larger sample size. VI. Conclusions Some economists have written eloquently of the revival of the study of religion by their fellow economists. Iannaccone [1998] is hopeful that the revival will soon bury the myths that homo economicus has no need of piety and that homo religiosus is a throw back to pre-rational times. Others have lauded the return of some economists to explaining the fundamental elements of human satisfaction and happiness and Frey and Stutzer (2002) also believe that this new or refound interest will stimulate their

9 8 interest in "issues such as good governance and social capital". Yet at the same time there has been very little interest in the relationship between happiness and religion among economists. This may reflect Tomes (1985) assessment that "economics is fundamentally atheistic". While it is no longer true for economics in general it still has been true for research on what leads to happiness. This paper is a contribution to filling that void. It finds that if we look at a large sample of individuals in Europe, their life satisfaction is positively related to measures of strong religious attachment in the sense of being willing to commit to attending religious services frequently. We have also shown that when we look at this sample and consider the hypothesis that people who identify their politics differently may have different reactions to income inequality, the conclusion that others reach that leftists suffer more is not strongly corroborated. The sample and our specification also throws into substantial doubt any conclusions that overall unemployment rates must be weighed more heavily than inflation rates in misery indices. Neither seems to have any significant effect on perceived satisfaction in our experiments that allow controls for religious phenomena. At the same time we do not desire to overreach. While it appears that church attendance increases perceived life satisfaction, it is at least conceivable that there are other explanations besides the efficacy of religious practice. One is that these attendance phenomena merely proxy for all kinds of social interaction that improve a person's state of mind. Argyle (1999) emphasizes the social support system religion provides. Another is that if church attendance is as we know it to be positively related to income, we may be picking up some of the effect of income on happiness that our imperfect income measure fails to capture. Since, however, Ellison does provide evidence that more telling measures of religious commitments in the U.S. beside religious attendance does increase life satisfaction and Stark (1972) shows that private religious devotions tend to be inversely related to income but positively related to attendance, we doubt that these are the whole explanations.

10 References 9 Alesina, Alberto, DiTella, Rafael and MacCulloch, Robert "Inequality & Happiness: Are Europeans and Americans Different?". National Bureau of Economic Research Working Paper # New York. Alesina, Alberto and La Farrara, Eliana "Preference for Redistribution in the Land of Opportunity". National Bureau of Economic Research Working Paper # New York. Argyle, Michael Causes and Correlates of Happiness. In Kahneman Daniel, Diener, Ed, and Schwartz, Norbert (Eds.), Well-Being: The Foundations of Hedonic Psychology New York. Russell Sage. Buttel, Frederick, Wilkening E. and Martinson, Oscar Ideology and Social Indicators of the Quality of Life. Social Indicators Research 4: Chiriboga, D. A Consistency in Adult Functioning: The Influence of Social Stress. Aging and Society. 2: Deininger, Klaus and Squire, Lyn " A New Data Set Measuring Inequlity". World Bank Economic Review. 10: DiTella, Rafael, MacCulloch, Robert and Oswald, Andrew "Preferences over Inflation & Unemployment Evidence from Surveys of Happiness". American Economic Review. 91: Ellison, Christopher "Religious Involvement & Subjective Well Being". Journal of Health and Social Behavior. 32: Freeman, Richard "Who Escapes? The Relation of Churchgoing & Other Background Factors to the Socioeconomic Performance of Black Male Youths from Inner City Tracts" in The Black Youth Employment Crisis. Richard Freeman and Harry Holzer eds. Chicago: University of Chicago Press. Frey, Bruno and Stutzer, Alois. 2002a. Happiness & Economics: How the Economy & Institutions Affect Human Well Being. Princeton: Princeton University Press. Frey, Bruno and Stutzer, Alois. 2002b. "What Economists Can Learn from Happiness Research?". Journal of Economic Literature. 40: Gardner, Jonathan and Oswald, Andrew "Does Money Buy Happiness? A Longitudinal Study Using Data on Windfalls". Working Paper, Warwick University Glaeser, Edward Measuring Trust. Quarterly Journal of Economics. 115: Glaeser, Edward and Sacerdote, Bruce "Education and the Demand for Happiness. Working Paper, Harvard University.

11 10 Hollander, Heinz "On the Validity of Utility Statements: Standard Theory versus Duesenberry's". Journal of Economic Behavior and Organization. 45: Hull, Brooks Religion Still Matters. The Journal of Economics. 24: Hull, Brooks and Bold, Frederick Preaching Matters: Replication and Extension. Journal of Economic Behavior and Organization. 27: Iannaccone, Lawrence "Introduction to the Economics of Religion". Journal of Economic Literature. 36: Interuniversity Consortium for Political & Economic Research Euro-Barometer Surveys. Ann Arbor: University of Michigan. Kahneman Daniel, Diener, Ed, and Schwartz, Norbert. (Eds.) (1999). Well-Being: The Foundations of Hedonic Psychology. New York. Russell Sage. Lipford, Jody, McCormick, Robert and Tollison, Robert "Preaching Matters". Journal of Economic Behavior and Organization. 21: Lowenstein, George "Because It Is There: The Challenge of Mountaineering for Utility Theory". Kyklos. 53: Nelson, Phillip and Greene, Kenneth Signaling Goodness: Social Rules and Public Choice. Ann Arbor. University of Michigan Press. Pahnke, W. H Drugs and Mysticism. International Journal of Parapsychology. 8: Sirgy, Joseph Materialism and Quality of Life. Social Indicators Research. 43: Smith, Adam. [1776]1981. An Inquiry into the Nature and Causes of the Wealth of Nations. Indiannapolis. Liberty Fund. Stark, Rodney "The Economics of Piety". In Gerald Thielbar and Saul Feldman (Eds.), Issues in Social Inequality Boston: Little Brown. Tomes, Nigel "The Effects of Religion and Denominations on Earnings & Returns to Human Capital". American Economic Review. 75: Veenhoven, Ruut Freedom and Happiness: A Comparative Study in Forty Four Nations in the Early 1990 s. In Ed Diener and Eunkook Suh (Eds.), Culture and Subjective Well Being Cambridge: MIT Press.

12 Endnotes For an insightful discussion of the role of individual autonomy and feelings of well being, see Loewenstein (1999). 2. First negatively among adults and then positively. 3. They attribute the different reactions to the greater mobility across income groups in the United States. Corroborating evidence for this hypothesis is presented in [Alesina and LaFarrara 2000]. 4. These included the experience of divorce, personal unemployment, bereavement and hospitalization coded from 0 to Coded from 0 to Coded from 1 to Coded from 1 to 8., 8. These are liberal protestant denominations, Mormon/Jehovah's Witnesses, non-denominational Protestant. The results on a similar regression explaining a happiness index are a bit different with only conservative Protestant, non-christian religions and Catholics having significant negative effects. 9. It is true that it is conceivable that causation may work in the reverse direction and that an individual's economic characteristics like high incomes may be caused by his high perceived well being and a person's religion. Characteristics like attendance might be affected by his state of mind. But it is by no means clear that this would result in our simple specification over-estimating the effect of say income and religious attendance. So Sirgy (1997) argues that those who prize material goods may have personalities that cause them to be less happy. It is also unclear whether say a positive frame of mind leads to more religious attendance. Religious attendance tends to increase in hard times and directly after events like September 11, 2001 in the United States. Also there is some good evidence that for religion the causation is from it to happiness rather than the reverse. So Chiriboga (1982) found that earlier church attendance predicted happiness positively fifteen years later for elderly individuals and Pahnke (1966) found that religious experience produce happiness later on. 10. We would also like to control for the political, economic and personal freedom enjoyed. This is because Loewenstein (1999) makes a strong a priori case for personal autonomy improving life satisfaction, because Frey and Stutzer (2002b) attest to an empirical connection and their work on Swiss cantons (2002a) bolster such an idea. Unfortunately, the only definitive ranking of countries by Veerhoven (2000) provides quantitative measures for only eleven of our countries and for only one year. It is interesting, however, that when for these eleven countries we explain the regression coefficients of the country dummies in Table 1, column 1 by the Veerhoven index of personal political and economic freedom, the coefficient for it is positive and significant with at statistic of 2.1, despite the very small sample size. So the country dummies may in part be a measure of freedom and explain why the coefficients for the Netherlands and Denmark are particularly high and for Portugal and Greece particularly low.

13 11. It can be argued that the religious variables may explain satisfaction because people who participate in social activities are more satisfied and that religion is just one such activity. For evidence along those lines see Glaeser and Sacerdote (2002) Glaeser (1998) shows that all kinds of social activity tend to fall with place size. 13. Though the coefficients for the Ginis are lower for the right than the neutral and the latter are lower than for the leftists, the results are not statistically significantly different.

14 Table 1. Variable Definitions and Summary Statistics of the Data 13 A. Dependent Variable: Life Satisfaction Category: With the life you lead, you are Count (Proportion) Count (Proportion) 1: Very Satisfied (0.272) 7323 (0.270) 2: Fairly Satisfied (0.537) (0.534) 3: Not Very Satisfied 38935(0.142) 4083 (0.151) 4: Not At All Satisfied 13382(0.049) 1211 (0.045) Total Count B. Independent Variables Variable: Definition Mean St. Dev. Mean St. Dev. GINI in % CPI in % rate of increase UR: unemployment rate in % RGDP: Real GDP per Capita in $ at the price levels and exchange rates of Unemployment=1, if (the respondent is) unemployed Selfemp=1 if self-employed Retired=1, if retired Student=1, if student HseWife=1, if housewife Male=1, if male Age in years AgeSq: Age squared Ed15_18=1, if years of education Ed19up=1, if Ed 19 year Married=1, if married Divorce=1, if divorced Separat=1, if separated Widowed=1, if widowed KidBg1=1, if 1 child aged between 8 and KidBg2=1, if 2 children aged between 8 and KidBg3p=1, if 3 or more children aged between 8 and Qtile2=1, if 2 nd income quartile Qtile3=1, if 3 rd income quartile Qtile4=1, if 4 th income quartile Religion: Catholic=1, if Catholic Protestant=1, if Protestant a Other=1, if other religion Orthodox=1, if Greek Orthodox Reference = no religion Church Attendance per week: Catholic*2=1, if twice or more and Catholic Catholic*1=1, if once and Catholic Protestant*2=1, if twice or more and Protestant Protestant*1=1, if once and Protestant Other*2=1, if twice or more and other religion Other*1=1, if once and other religion Orthodox*2=1, if twice or more and Greek Orthodox Orthodox*1=1, if once and Greek Orthodox OftPol=1, if frequent political discussion BigTown=1, if community size is largest third Left=1, if left political view, scoring 1 4 (in a scale of 1 10) Right=1, if right political view scoring Observations Includes Gereformeerd, Church of Scotland, and Non-conformist church.

15 Table 2. The Estimated Ordered Logit Model for Life Satisfaction 14 (1) (2) Variable Estimate St. Err Estimate St. Err GINI ** CPI UR RGDP Unemp ** ** Selfemp ** Retired ** Student * HseWife ** Male ** ** Age ** ** AgeSq ** ** Ed15_ ** * Ed19up ** ** Married ** ** Divorce ** ** Separat ** ** Widowed ** ** KidBg ** KidBg ** KidBg3p ** ** Qtile ** ** Qtile ** ** Qtile ** ** Catholic Protestant Other Orthodox Catholic* ** Catholic* ** Protestant* ** Protestant* ** Other* ** Other* ** Orthodox* ** Orthodox* ** OftPol BigTown ** Left ** Right Intercept ** Inter ** ** Inter ** ** Obs log L * stands for significance at 10% and ** for significance at 5%.

16 Table 3. Life Satisfaction and Political View: Left Right Sub-samples 15 Left Neutral Right Variable Estimate St. Err Estimate St. Err Estimate St. Err GINI ** ** ** CPI ** UR RGDP Unemp ** ** ** Selfemp Retired Student HseWife Male ** ** Age ** ** ** AgeSq ** ** ** Ed15_ ** Ed19up ** Married ** ** ** Divorce ** ** ** Separat ** ** ** Widowed ** ** KidBg KidBg * ** KidBg3p ** ** Qtile ** ** ** Qtile ** ** ** Qtile ** ** ** Catholic Protestant Other Orthodox * Catholic* ** ** ** Catholic* ** ** Protestant* * ** ** Protestant* ** ** ** Other* ** ** ** Other* ** Orthodox* ** Orthodox* * * OftPol BigTown ** ** ** Intercept Inter ** ** ** Inter ** ** ** Obs log L