Date: 08/31/2015 Impact of Unemployment on Demonstrated Happiness in the United States Mohammad Sameem 1 and Pavlo Buryi 2 ABSTRACT Among the many documented consequences of unemployment is a higher level of selfreported unhappiness. The present work considers instead the level of demonstrated happiness and unhappiness within groups, the latter proxied by the conditional probability of suicide within groups facing a higher unemployment rate and those without. Contrary to the literature, the ones with higher unemployment rate feel happy, in these terms, and actually have slightly lower rates of suicide than those with the job, based on a recent U.S. data. Keywords: unemployment, happiness, education, suicide. JEL Classification: I1, I2 1 Corresponding author, Economics Department, Southern Illinois University, Carbondale, IL USA. E-mail: sediq.sameem@siu.edu 2 Economics Department, Southern Illinois University, Carbondale, IL, USA. E-mail: pasha@siu.edu
2 1. INTRODUCTION During the year 2004, more than 32,000 suicides took place in the United States 3. Of this total, more than 23,000 cases are included in this research. What could have driven the people to commit suicide? One of the primary causes of attempting such a behavior is an extreme level of unhappiness. Both higher and longer unemployment could lead to higher level of anxiety, hence, cause suicide. However, empirically the scenario has not been so conclusive. On one hand, a counter-cyclical pattern of suicide rate has been found in the United States (Ruhm, 2000), Denmark (Mortensen et al., 2000), Italy (Preti and Miotto, 1999) and Sweden (Johansson and Sundquist, 1997). On the other hand, a procyclical pattern has been found in Mexico (Gonzalez and Quast, 2010), Finland (Hintikka et al., 1999) and many other countries (Milner et al., 2012) whereas some studies find both scenarios (Garcy and Vagero, 2013) or remain inconclusive (Chen et al., 2010). In addition, the preexisting health issues, in general, and mental issues, in particular, may further confound this one-way causal relationship between unemployment and suicide because poorer health can also be a factor in the unemployment. A common approach in this literature on unemployment-suicide nexus is that the proxy used for happiness is a self-reported rather than a demonstrated one. This study instead considers the level of demonstrated happiness and unhappiness within groups, the latter proxied by the probability of suicide within groups facing higher unemployment and those without. Hamermesh and Soss s (1974) built a theoretical model that suggests that individuals take their own lives when the discounted expected lifetime utility remaining to them falls below some threshold level. The model then predicts that good political institutions and governance structures make people better off. Antonio Rodriguez Andres and Justina A. V. Fischer (2008) examines the effects of political institutions and governance structure on suicide using a balanced panel for 26 Swiss states (cantons) over the period 1980 1998. They find that stronger popular rights and more fiscal decentralization reduce suicide, while more local autonomy increases it. Daly, Wilson, and Johnson (2007) also proposes using data on suicide as a measurement of outcome based wellbeing, and find strong evidence that reference-group income negatively affects suicide risk. Mary Daly and Wilson (2008) compares and contrast the empirical patterns of subjective wellbeing and suicide data. They find that the two have very little in common in aggregate data (time series and cross-sectional), but have a very strong relationship in terms of their determinants in individual-level, multivariate regressions. Using recent U.S. data on unemployment and suicide rates we find that individuals who live in the states with higher rates of unemployment are demonstratively happier in the sense of having reduced tendency towards committing suicide compared to those who live in the states with lower rates of unemployment. Furthermore, consistent with Buryi and Gilbert (2014), our results suggest a higher tendency towards suicide among the individuals with higher levels of education. Possible explanations for such results include lower opportunity costs of joining social, health, and exercise clubs, hence, lower level of anxiety, and lower returns to education after the huge burden of student loans. 3 Data source: Center for Disease Control and Protection (CDC), U.S. Department of Health and Human Services
3 2. DATA AND METHODOLOGY Our main concern is testing the hypothesis that unemployment has negative effects on happiness that might result in suicide. To test this hypothesis, we use the 2004 U.S. Vital Statistics Multiple Cause of Death Data 4. This dataset covers all individuals died during 2004 in the U.S. The dataset includes information about individuals age, sex, race, marital status, education status, state of death occurrence, and so on. For our model, the demonstrated happiness is measured by the conditional probability of committing suicide. The dependent variable (suicide) equals 0 if the person is completely happy and 1 if the person is completely unhappy. The happier the person the less likely s/he is to commit suicide. The independent variable of interest is the ascending rank of states based on unemployment during 2004. For our estimation we use logit and probit models because these models allow us to compute the conditional probabilities of committing suicide as the states ranging from lower unemployment rate to a higher one while controlling for age, sex, race, marital status, the region of occurrence, and education level. Our sample is limited to only white and black individuals who have at least completed high school and died within any of the four regions of the U.S from either natural causes or suicide. In order to eliminate the possibility of finding any residual effect of unemployment on dying from suicide, we restrict our sources of death to the two categories of natural causes and suicide only. On the premise that unemployment could enhance happiness, people would be less likely to commit suicide in the states with higher rates of unemployment, hence, increasing the chances of dying from natural causes. Our model of estimation is: P i = α + βurrank i + γedu i + δx i + ε i (1) P i equals 1 when a person committed suicide and 0 when a person died from a natural cause, both URrank and Edu 5 are ordinal variables indicating the ranking of states based on unemployment and education in years, respectively, and X includes all other variables controlling for sex, race, age, marital status of the individuals and the region where the death occurred. Table 1 reports our estimation results. The coefficients indicate log odds or simply likelihood of committing suicide for a one unit change in the explanatory variables. Controlling for everything else, our findings indicate that there is a lower tendency towards committing suicide, hence, higher level of happiness in the states having higher unemployment rate. The results are consistent in both models of logit and probit. Several explanations are in order. First, opportunity cost of health-intensive activities is much lower during the times of recession when unemployment rate is higher. Therefore, people can substitute into such activities that might improve their overall health, in general, and mental health, in particular. Second, an important psychological aspect of unemployment is a social stigma, unemployed individuals may feel inferior to those employed in good positions. This effect becomes more dominant during the period of lower unemployment as compared to that of higher unemployment when the job loss becomes a norm rather than deviation from norm. Therefore, suicide rate might be lower in states where most of the people are unemployed. 4 We select 2004 since it is the latest year for which the state of death occurrence is reported by the States. 5 Edu variable corresponds to educ1989 variable in the Multiple Cause of Death Data.
4 Table 1. Logit and Probit Estimates (1) (2) (3) (4) Suicide Logit Probit UR Rank -0.011*** -0.016*** -0.005*** -0.007*** (-16.772) (-21.703) (-16.200) (-21.230) Education 0.072*** 0.030*** (14.879) (13.485) Male 0.935*** 0.948*** 0.440*** 0.444*** (54.388) (47.630) (56.759) (49.355) White 1.327*** 1.373*** 0.648*** 0.672*** (43.601) (39.324) (46.610) (42.116) Married 0.027 0.092*** -0.061*** -0.029** (1.320) (3.847) (-6.104) (-2.475) Divorced 0.398*** 0.491*** 0.104*** 0.145*** (17.938) (18.973) (9.454) (11.292) Widowed 0.381*** 0.505*** 0.166*** 0.223*** (10.683) (12.095) (10.967) (12.556) Age -0.098*** -0.101*** -0.045*** -0.046*** (-176.78) (-155.78) (-173.91) (-152.90) Northeast -1.124*** -0.937*** -0.549*** -0.455*** (-46.073) (-28.022) (-48.316) (-28.978) Midwest -0.759*** -0.561*** -0.381*** -0.277*** (-35.805) (-20.221) (-37.620) (-20.730) South -0.737*** -0.572*** -0.360*** -0.275*** (-37.016) (-22.178) (-37.703) (-22.085) Constant 99.276*** 101.253*** 44.960*** 45.848*** (172.769) (151.613) (168.719) (147.683) N 992,989 742,782 992,989 742,782 Pseudo R 2 0.307 0.307 0.305 0.304 Notes: t-statistics are in parentheses. *** p<0.01, ** p<0.05, * p<0.1 Consistent with previous literature, men, whites, and individuals with higher level of education are more likely to take their own lives as compared to women, blacks, and least educated people. In comparison to singles and youngsters, married, divorced and widowed are more likely (more reliant on jobs) and older people (less reliant) are less likely to commit suicide. We also control for each of the four regions; Northeast, Midwest, South, and West where the last region seems to be the one where people are more likely to commit suicide, which could be explained by the fact that people in the former three regions are relatively more godfearing church-goers and do not get so uptight about failed ambitions. To make reasonable inferences about the magnitude of the effect of these variables on the probability of committing suicide one needs to understand the marginal effects, which could be computed either at means or at any specific point. Since most of our explanatory variables are either indicator or ordinal variables computing marginal effects at means does not make much sense, hence, we calculate these effect at every point since a one unit increase in the explanatory
5 variable could have different effect on the outcome variable depending on the initial values of the explanatory variables. Figure 1 exhibits conditional probabilities of suicide in the states..032 Marginal Effects with 95% CIs.02 Pr(Suicide).022.024.026.028.03 0 10 20 30 40 50 UR Rank (Ascending) Figure 1 Marginal Effects of Unemployment on Demonstrated Happiness Figure 1 suggests not only that higher unemployment rate has negative effect on the probability of committing suicide but also it is lower in the states with higher unemployment rate. For instance, moving from a state ranked first, in ascending order, in terms of lower unemployment rate to a state ranked as high as 49 th decreases the probability of committing suicide from 3 percent to 2 percent, which makes quite a significant difference in real world. 3. CONCLUSION Evaluating happiness in terms of likelihood of committing suicide, the present work suggests that living in a state of the U.S. with a higher unemployment rate increases the happiness of the individuals, in these terms, and actually reduces the probability of committing suicide. Although we cannot make inferences about causation, the fact that unemployment and demonstrated happiness are positively correlated could be useful if individual happiness, rather than economic outcomes, is the focus of government policies.
6 4. REFERENCES Buryi, P. and S. Gilbert, 2014, "Effects of College Education on Demonstrated Happiness in the United States" Applied Economic Letters, Vol. 21, No. 18, pp. 1253-1256 Chen, Y-Y., PSF. Yip, C. Lee, H-F Fan and K-W Fu, 2010, "Economic Fluctuations and Suicide: A Comparison of Taiwan and Hong Kong" Social Science and Medicine, Vol. 71, pp. 2083-2090 Daly, M.C., Daniel Wilson, and Norman Johnson. Federal Reserve Bank of San Francisco Working Paper 2007-12 Daly, M.C., Daniel Wilson, and Norman Johnson. Federal Reserve Bank of San Francisco Working Paper 2008-19 Garcy, A. and D. Vagero, 2013, "Unemployment and Suicide During and After a Deep Recession: A Longitudinal Study of 3.4 Million Swedish Men and Women" American Journal of Public Health, Vol. 103, pp. 1031-1038 Gonzalez, F. and T. Quast, 2010, "Mortality and Business Cycles by Level of Development: Evidence from Mexico" Social Science and Medicine, Vol. 71, pp. 2066-2073 Hamermesh, D, and Soss, N., 1974 An Economic Theory of Suicide Journal of Political Economy Vol. 82, No. 1, pp. 83-98 Hintikka, J., P. Saarinen and H. Viinamaki, 1999, "Suicide Mortality in Finland during an Economic Cycle, 1985-1995" Scandinavian Journal of Public Health, Vol. 27, No. 2, pp. 85-88 Johansson S.E. and J. Sundquist, 1997, "Unemployment is an Important Risk Factor for Suicide in Contemporary Sweden: an 11-Year Follow-up Study of a Cross-sectional Sample of 37789 People" Public Health, Vol. 111, pp. 41-45 Milner, A., R. McClure and D. Leo, 2012, "Socio-economic Determinants of Suicide: An Ecological Analysis of 35 Countries" Social Psychiatry and Psychiatric Epidemiology, Vol. 47, pp. 19-27 Mortensen P., E. Agerbo, T. Erickson, P. Qin and P. Westergaard-Nielsen, 2000, "Psychiatric Illness and Risk Factors for Suicide in Denmark" Lancet, Vol. 355, No. 9197, pp. 9-12 Preti, A. and P. Miotto, 1999, "Suicide and Unemployment in Italy, 1982-1994" Journal of Epidemiol Community Health, Vol. 53, pp. 694-701 Rodriguez Andres, A., and Fischer, J. 2008, Political institutions and suicide: A regional analysis of Switzerland, Thurgau Institute of Economics and Department of Economics at the University of Konstanz Research Paper Series, Paper N33 Ruhm, C., 2000, "Are Recessions Good For Your Health?" Quarterly Journal of Economics, Vol. 115, No. 2, pp. 617-650