Abstract
The link between education and health is one of the most robust empirical relationships in the social sciences. However, little research has examined the effects of educational quality on health outcomes. We estimate the long run relationship between health behaviors and graduating from a selective college in the 1960s using the Wisconsin Longitudinal Study, which has tracked siblings for over 50 years. Importantly, we control for measures of health endowments, ability, and time preferences before college enrollment as well as shared family and environmental factors. We find large effects of college selectivity on reducing overweight for individuals in their 60s.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
For further details about the representativeness of the WLS or more information about the sample design, see Fletcher and Frisvold (2012) or Wisconsin Longitudinal Study (2006). In general, further details about the data and the construction of our analysis sample are available in our working paper (Fletcher and Frisvold 2012).
Barron’s Profile of American Colleges provides an overview of 4-year colleges in the United States that includes their admissions standards, expenses, and programs of study. We use the 1969 edition of this resource, which provides the earliest measures of college quality, to our knowledge. Because most survey respondents attended college between 1958 and 1963, there is likely to be measurement error in the college quality variables. However, due to the stability of college rankings, the measurement error is likely minimal.
The information used to construct these selectivity categories is not available for all colleges; thus, we focus our analysis on the selectivity categories: (1) most competitive, (2) highly competitive, (3) very competitive, (4) competitive, (5) less competitive, and (6) non-competitive.
Examples of colleges in the very competitive category include University of Wisconsin, University of Michigan, and University of Minnesota.
As shown in Table 1, for the sample of all college graduates (n = 3,210), only 3.6 % graduated from a highly competitive college. As shown in Table 2, among the sibling sample (n = 530), only 3.2 % graduated from a highly competitive college. Thus, the number of individuals graduating from a highly competitive college is very small, and, for this reason, we do not focus our analysis on individuals graduating from a highly competitive college. As shown in Table 1, for the sample of all college graduates, 64.2 % graduated from a competitive college. As shown in Table 2, among the sibling sample, 64.9 % graduated from a competitive college. The selection criteria for schools to be classified as competitive is a median SAT score of 900–1,100 and students should have an average GPA of B− or better and be in the top half of their high school class. As would be expected because the cutoff for a competitive college is not very selective, the magnitudes of the estimates are smaller and most are not statistically significant. Thus, because the highly competitive colleges are so selective as to only include a few people in the sample and the competitive colleges are not very selective and include over half the sample, we focus our analysis on graduating from a very selective college.
“Appendix” Table 1 displays the characteristics of WLS respondents who were not included in this analysis. As expected, the individuals who graduated from college during this time period are relatively advantaged. Compared to the full sample of respondents, college graduates have mothers and fathers with more years of schooling completed, were raised in smaller families with higher incomes, and have higher IQ scores. College graduates with missing health data have similar pre-college characteristics to the college graduates included in Table 1.
We also assign siblings with missing values the value of the non-missing sibling for living with both parents and assign sibling pairs with missing values the imputed value of the graduate for mother’s education, father’s education, and family income; these missing values will difference out in the fixed effects model.
Graduating from a very competitive college is not correlated with whether the respondent is included in this analysis sample.
The correlation between graduating from a very competitive college and mortality by 2004 is −0.01 and is not statistically different from zero.
Only the name of the college that the individual graduated from is available for both the original WLS respondents and their siblings. Thus, we focus on the college selectivity of the college that individuals graduated from, as opposed to the college that individuals initially attended. In “Appendix” Table 3, we show the results from OLS and linear probability models of the relationship between attending a very competitive college and health behaviors for the sample of original WLS respondents. These results show that there is not a statistically significant relationship between attending a selective college and later health behaviors; thus, based on our results below, college graduation is the more relevant margin for influencing health behaviors. We also show that our results are very similar if we compare the WLS respondents with a sibling in the sample with WLS respondents without a sibling in the sample. We will return to these results below.
Although this effective sample size is less than ideal, we note that it is comparable to many other studies using sibling comparisons and the sample size should influence the precision but not the magnitude of the estimates (Garces et al. 2002).
These measures of childhood health are derived from questions asked of respondents in the latest survey wave, which introduces the possibility of substantial recall error. On the other hand, significant events, such as illness for at least 1 month, are less likely to be subject to recall error (Garces et al. 2002). The most commonly reported illness is infectious and parasitic diseases. In the WLS, childhood measures of health behaviors are unavailable. It could be possible that individuals are overweight in childhood but do not report poor or fair childhood health or a prolonged childhood illness and that childhood overweight is related to both adult overweight and college selectivity. However, Kaestner and Grossman (2009) and Fletcher and Lehrer (2009) suggest that there is no relationship between childhood overweight and educational achievement.
Further information about high school experiences is available for the original WLS respondents, but not the siblings. We examine the influence of the extent to which the respondent discussed their future (post-high school) plans with teachers, counselors, and parents, as an alternative proxy for future expectations, and whether the respondent participated in high school sports on the estimated impact of college selectivity using the sample of original WLS respondents only. The inclusion of these additional variables does not influence the results.
As noted in Fletcher and Frisvold (2009), a limitation of the fixed effects strategy is the exaggeration of the influence of measurement error, which could bias the estimate of δ2 towards zero. Additionally, while the fixed effects strategy is implemented to reduce endogenous variation in college attendance, exogenous variation may also be reduced (Bound and Solon 1999). Further, sibling spillovers, where the behavior of one sibling is influenced by the other, would bias the estimates towards zero.
To further examine this possibility, we include personality characteristics (extraversion, agreeableness, conscientiousness, neuroticism, and openness) as additional covariates and find that our results are robust to including these characteristics. A caveat to the importance of this robustness check is that the personality characteristics are measured during adulthood in 1993 and Roberts et al. (2006) conclude that these personality traits are not fixed throughout adulthood (see also Almlund et al. 2011). These results also suggest that changes in personality are not a mechanism through which graduating from a very selective college influences health behaviors.
The results are qualitatively similar if BMI is measured in levels.
As a way to provide suggestive evidence about this potential mechanism, we estimated models that adjusted for individual inheritance. Our results are robust to this specification.
In Fletcher and Frisvold (2012), we examine the impact of graduating from a selective college on total household income with a sibling fixed effects specification. Consistent with the existing literature, our results show that college quality increases total household income, but only in 2004. The estimated increase in total household income is 36 % at age 65 in 2004. We also examine whether graduating from a selective college increases graduate school attendance and find that graduating from a selective college increases the probability of graduate school attendance by 16 % points and increase the years of schooling completed by 0.36 years.
Since the average age is 64 years in 2004, a potential implication from Table 7 is that graduating from a very selective college induces individuals to work longer. Similar to total household income, we find that controlling for employment status has little impact on the results.
Similar changes occur when controlling for years of schooling completed instead of a dummy variable for graduate school attendance or when controlling for years of schooling completed and graduate school attendance.
In addition to marital status, spouse’s characteristics may also influence health behaviors. Controlling for marital status and spouse’s education level does not influence the estimates of college selectivity. Information about the selectivity of the college attended by spouses is unavailable and remains a potential mechanism that could explain these results.
In contrast, Cutler and Lleras-Muney (2010) present suggestive evidence that they can “explain” up to 30 % of the educational quantity gradient and health behaviors (smoking, heavy drinking, and obesity) with measures of resources (e.g. income) and up to 30 % with measures of cognitive ability. However, they are unable to examine the relationship between educational quality and health behaviors.
We calculate occupation-specific obesity rates within-sample so that the measure represents the proportion of WLS respondents (excluding the focal individual) who are obese in each occupation. We also use respondents’ reports of whether they work with smokers at their current job (or last job for those who are retired).
An additional potential issue is that the inclusion of possibly endogenous variables could bias the college quality estimate; however, based on the results shown in Table 9, this possibility does not seem likely in this context.
Gelbach points out that sequential addition can lead to biased estimates of the impact when examining the robustness of the impact to the inclusion of additional covariates. He develops an alternative approach that provides consistent estimates of the influence of additional variables on the relationship of interest, but notes that this method may not be appropriate when the effect of the primary variable of interest operates through the additional variables. Gelbach also has created Stata code to implement his estimator “b1x2.ado”, which we use: http://gelbach.eller.arizona.edu/papers/b1x2/.
As we discussed above, when we examined the effects of attending a selective college rather than focusing on graduating from a selection college on health behaviors for the WLS respondents, we found smaller and insignificant effects. Although these results were not able to use sibling fixed effects because the survey does not contain attendance information of the siblings of the WLS respondents, we posit that the differential results for attendance versus graduation could suggest a limited role for some classes of mechanisms, such as exposure to college peers and changes in preferences/knowledge. That is, a discrete effect of college that appears at college graduation but not before may be inconsistent with the (presumably continuous) effects of college peers, preference formation, and knowledge acquisition during the college years. However, we note that this argument is necessarily speculative.
Although a potential limitation is the external validity of this sample, which is drawn from a single state in 1957, this sample is broadly representative of white, non-Hispanic adults with at least a high school degree who were born during this period. Additionally, if the influences of college quality do not vary substantially across states, then results from this sample would be similar to results from a nationally representative sample (Jencks et al. 1983). Further, we view the wealth of information available in the WLS that is not available in nationally representative data sets as a benefit that exceeds the costs of any reduced external validity.
References
Abadie, A., & Imbens, G. (2002). Simple and bias-corrected matching estimators for average treatment effects. NBER Technical Working Paper 283.
Aizer, A., & Stroud, L. (2010). Education, knowledge and the evolution of disparities in health. National Bureau of Economic Research Working Paper 15840.
Albouy, V., & Lequien, L. (2009). Does compulsory education lower mortality? Journal of Health Economics, 28(1), 155–168.
Almlund, M., Duckworth, A. L., Heckman, J., & Kautz, T. (2011). Personality psychology and economics. IZA Discussion Paper 5500.
Arendt, J. (2005). Does education cause better health? A panel data analysis using school reforms for identification. Economics of Education Review, 24, 149–160.
Behrman, J., Rosenzweig, M., & Taubman, P. (1996). College choice and wages: Estimates using data on female twins. Review of Economics and Statistics, 78(4), 672–685.
Black, D., & Smith, J. (2004). How robust is the evidence on the effects of college quality? Evidence from matching. Journal of Econometrics, 121, 99–124.
Bound, J., & Solon, G. (1999). Double trouble: On the value of twins-based estimation of the return to schooling. Economics of Education Review, 18, 169–182.
Case, A., Fertig, A., & Paxson, C. (2005). The lasting impact of childhood health and circumstance. Journal of Health Economics, 24, 365–389.
Card, D. (1999). The Causal Effect of Education on Earnings. In O. Ashenfelter & D. Card (Eds.), Handbook of Labor Economics (pp. 1802–1863). Amsterdam: Elsevier.
Clark, D., & Royer, H. (2010). The effect of education on adult health and mortality: Evidence from Britain. NBER Working Paper 16013.
Cockerham, W. C., Rutten, A., & Abel, T. (1997). Conceptualizing contemporary health lifestyles: Moving beyond Weber. Sociological Quarterly, 38, 601–622.
Conti, G., Heckman, J., & Urzua, S. (2010). The education-health gradient. American Economic Review, Papers and Proceedings, 100(2), 234–238.
Crump, R. K., Hotz, V. J., Imbens, G. W., & Mitnik, O. A. (2006). Moving the Goalposts: Addressing Limited Overlap in the Estimation of Average Treatment Effects by Changing the Estimand. NBER Technical Working Paper 330.
Cutler, D., & Lleras-Muney, A. (2008). Education and health: Evaluating theories and evidence. In R. F. Schoeni, J. S. House, G. Kaplan, & H. Pollack (Eds.), Making Americans healthier: Social and economics policy as health policy. New York: Russell Sage Foundation.
Cutler, D., & Lleras-Muney, A. (2010). Understanding differences in health behaviors by education. Journal of Health Economics, 29(1), 1–28.
Dale, S., & Krueger, A. (2002). Estimating the payoff to attending a more selective college: An application of selection on observables and unobservables. Quarterly Journal of Economics, 117(4), 1491–1527.
de Walque, D. (2007). Does Education Affect Smoking Behaviors? Evidence Using the Vietnam Draft as an Instrument for College Education. Journal of Health Economics, 26, 877–895.
Fine, B. (1969). Barron’s profiles of American colleges. Woodbury, NY: Barron’s Educational Series.
Flegal, K. M., Graubard, B. I., Williamson, D. F., & Gail, M. H. (2005). Excess deaths associated with underweight, overweight, and obesity. Journal of the American Medical Association, 293, 1861–1867.
Fletcher, J. M., & Frisvold, D. E. (2009). Higher education and health investments: Does more schooling affect preventive care use? Journal of Human Capital, 3(2), 144–176.
Fletcher, J. M., & Frisvold, D. E. (2011). College selectivity and young adult health behaviors. Economics of Education Review, 30, 826–837.
Fletcher, J. M., & Frisvold, D. E. (2012). The long run health returns to college quality. Federal Reserve Bank of Atlanta Center for Human Capital Studies Working Paper 2012-1.
Fletcher, J. M., & Lehrer, S. F. (2009). The effects of adolescent health on educational outcomes: Causal evidence using ‘genetic lotteries’ between siblings. Forum for Health Economics & Policy, 12(2), article 8.
Frisvold, D., & Golberstein, E. (2011a). School quality and the education–health relationship: Evidence from blacks in segregated schools. Journal of Health Economics, 30, 1232–1245.
Frisvold, D., & Golberstein, E. (2011b). The effect of school quality on black-white health differences: Evidence from southern segregated schools. Federal Reserve Bank of Atlanta Center for Human Capital Studies Working Paper 2011-1.
Fuchs, V. R. (1982). Time Preference and Health: An Exploratory Study. In V. Fuchs (Ed.), Economic Aspects of Health. Chicago: University of Chicago Press.
Garces, E., Thomas, D., & Currie, J. (2002). Longer-term effects of head start. American Economic Review, 9(4), 999–1012.
Gelbach, J. (2009). When do covariates matter? And which ones, and how much? UA Department of Economics Working Paper #09-07. http://gelbach.eller.arizona.edu/~gelbach/papers/.
Grimard, F., & Parent, D. (2007). Education and smoking: Were Vietnam war draft avoiders also more likely to avoid smoking? Journal of Health Economics, 26(5), 896–926.
Grossman, M. (1972). On the concept of health capital and the demand for health. Journal of Political Economy, 80(2), 223–255.
Grossman, M., & Kaestner, R. (1997). Effects of education on health. In J. Behrman & N. Stancey (Eds.), The social benefits of education. Ann Arbor: The University of Michigan Press.
Hoekstra, M. (2009). The effect of attending the flagship state university on earnings: A discontinuity-based approach. Review of Economics and Statistics, 91(4), 717–724.
Hoxby, C. M. (2009). The changing selectivity of American colleges. National Bureau of Economic Research Working Paper 15446.
Ishida, H., Spilerman, S., & Su, K. H. (1997). Educational credentials and promotions chances in Japanese and American organizations. American Sociological Review, 62, 866–882.
Jencks, C., Crouse, J., & Mueser, P. (1983). The Wisconsin model of status attainment: A national replication with improved measures of ability and aspiration. Sociology of Education, 56, 3–19.
Johnson, R. (2010). Long-run impacts of school desegregation and school quality on adult attainments. University of California, Berkeley Working Paper.
Jones, A. M., Rice, N., & Dias, P. R. (2011). Long-term effects of school quality on health and lifestyle: Evidence from comprehensive schooling reforms in England. Journal of Human Capital, 5(3), 342–376.
Kaestner, R., & Grossman, M. (2009). Effects of weight on children’s educational achievement. Economics of Education Review, 28(6), 651–661.
Kenkel, D. (1991). Health behavior, health knowledge, and education. Journal of Political Economy, 99(2), 287–305.
Lleras-Muney, A. (2005). The relationship between education and adult mortality in the US. The Review of Economic Studies, 72, 189–221.
Long, M. C. (2008). College quality and early adult outcomes. Economics of Education Review, 27(5), 588–602.
MacInnis, B. (2009) Returns to school quality on elderly cognition and cognitive aging. Working Paper.
Oreopoulos, P. (2007). Do dropouts drop out too soon? Wealth, health, and happiness from compulsory schooling. Journal of Public Economics, 91(11–12), 2213–2229.
Price, J. (2008). Parent–child quality time: Does birth order matter? Journal of Human Resources, 43(1), 240–265.
Roberts, B. W., Walton, K. E., & Viechtbauer, W. (2006). Patterns of mean-level change in personality traits across the life course: A meta-analysis of longitudinal studies. Psychological Bulletin, 132(1), 1–25.
Rosenbaum, J. E. (1984). Career mobility in a corporate hierarchy. New York: Academic Press.
Ross, C., & Mirowsky, J. (1999). Refining the association between education and health: The effects of quantity, credential, and selectivity. Demography, 36(4), 445–460.
Sander, W. (1995). Schooling and quitting smoking. Review of Economics and Statistics, 77(1), 191–199.
Sturm, R., & Wells, K. B. (2001). Does obesity contribute as much to morbidity as poverty or smoking? Public Health, 115, 229–235.
van Kippersluis, H., O’Donnell, O., & van Doorslaer, E. (2011). Long run returns to education: Does schooling lead to an extended old age? Journal of Human Resources, 46(4), 695–721.
Webbink, D., Martin, N. G., & Visscher, P. M. (2010). Does education reduce the probability of being overweight? Journal of Health Economics, 29(1), 29–38.
Wisconsin Longitudinal Study. (2006). Wisconsin Longitudinal Study Handbook. Available at http://www.ssc.wisc.edu/wlsresearch/documentation/handbook/WLS_Handbook.pdf.
Wisconsin Longitudinal Study. (2009). Retention and response rates of Wisconsin Longitudinal Study 1957–2005. Available at http://www.ssc.wisc.edu/wlsresearch/documentation/retention/cor1004_retention.pdf.
Wolfe, B., & Behrman, J. (1987). Women’s schooling and children’s health: Are the effects robust with adult sibling control for the women’s childhood background? Journal of Health Economics, 6(3), 239–254.
Acknowledgments
This research was supported by the WLS pilot grant program (Fletcher and Frisvold), the National Institute on Aging (R01 AG027045, Fletcher), the Emory Global Health Institute (Frisvold), and the Robert Wood Johnson Foundation (Fletcher and Frisvold). The authors thank Griffin Edwards and John Zimmerman for excellent research assistance and participants at the Academy Health, American Society of Health Economists, Annual Meeting on the Economics of Risky Behaviors, APPAM, and Southern Economic Association conferences, Dhaval Dave, Ezra Golberstein, Bo MacInnis, Dave Marcotte, and Jim Marton for helpful comments. This research uses data from the WLS of the University of Wisconsin-Madison. Since 1991, the WLS has been supported principally by the National Institute on Aging (AG-9775 and AG-21079), with additional support from the Vilas Estate Trust, the National Science Foundation, the Spencer Foundation, and the Graduate School of the University of Wisconsin-Madison. A public use file of data from the Wisconsin Longitudinal Study is available from the Wisconsin Longitudinal Study, University of Wisconsin-Madison, 1180 Observatory Drive, Madison, Wisconsin 53706 and at http://www.ssc.wisc.edu/wlsresearch/data/. The opinions expressed herein are those of the authors. In particular, the authors thank Carol Roan for assistance with the restricted college data.
Author information
Authors and Affiliations
Corresponding author
Appendix
Appendix
See Tables 8, 9, 10, 11 and 12.
Rights and permissions
About this article
Cite this article
Fletcher, J.M., Frisvold, D.E. The long run health returns to college quality. Rev Econ Household 12, 295–325 (2014). https://doi.org/10.1007/s11150-012-9150-0
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11150-012-9150-0