Over the past decade, there has been increasing interest in expanding access to college education in the US. Against a backdrop of rising tuition fees and falling college enrolment rates, states such as Oregon and Tennessee have implemented policies to improve the affordability of, or eliminate tuition fees for, two-year colleges.1 New York made state college or university education free for students in families earning up to $125,000.2 Several Democratic presidential candidates running for the 2020 election have discussed or released proposals to improve college affordability.3 An important and relevant question is: what are the causal impacts of a college education on future employment, earnings, and wellbeing? While many studies have examined labour market outcomes (Oreopoulos and Petronijevic 2013), fewer have looked at the impacts of college education on health-related behaviours and health outcomes.
Social science researchers have frequently studied the relationship between education and health, identifying a positive gradient between educational attainment and beneficial health behaviours and outcomes (Cutler and Lleras-Muney 2010). Economic theory predicts that there may be channels through which the relationship is causal (Grossman 1972, 2000). If education raises a person’s wages and improves their ability to process health-related information, they will be better equipped to consume higher quality and more appropriate health care, nutrition, and exercise. With the prospect of a longer and healthier life, they may also have greater incentives to invest in their health while young. However, there are other ways in which education and health are related – for example, negative health impacts in childhood may reduce educational attainment, or individuals may differ in their long-term outlook, thereby influencing investments in both education and health.
Most previous literature that considers the effects of education on health uses quasi-experiments that operate on margins of educational attainment prior to college (Galama et al. 2018). Others consider a specific sub-population, such as veterans of the Vietnam War (Buckles et al. 2013, De Walque 2007, Grimard and Parent 2007).4 Relatively few studies have reported causal effects of a college education on health outcomes for the general population.
Other than our own work, we are aware of only one other article that leverages measures of college access in order to identify the effects of college on health among the general population, and that is for Germany (Kamhöfer et al. 2018). We adopt a strategy closely related to that implemented by Currie and Moretti (2003), which counts the number of two- and four-year colleges per capita in each state and year (Cowan and Tefft 2020). We refer to this ratio as ‘college access’. When this omnibus measure is higher, it generally signifies that geographical proximity to an institution rises, the likelihood of admittance to college increases, and/or the pecuniary cost of college falls (all of which would boost the likelihood of enrolment).
Using reduced-form cohort-year-state fixed effects regression models, we study how college access directly affects college education, health outcomes such as self-reported health and health conditions, and health-related behaviours such as alcohol consumption, smoking, and exercise. Using college access as an instrumental variable in two-stage least squares (2SLS) regressions, we explore how increased years of education affects these health outcomes. We also study the heterogeneity of impacts among gender and racial subgroups.
Our study contributes to policy discussions about expanding college access and to the economics literature studying the relationship between education and health. It is among the first to study the effects of college access on health outcomes for the general population and for important population subgroups. Additionally, it is among the first to identify the causal effect of education on health, on the margin of college education.
Data and empirical strategy
We obtained the number of two- and four-year colleges by state in the US between 1960-1996 from Currie and Moretti (2003). We defined college access, for each state and year, as this number divided by the population of 18-22 year-olds in that state and year. We then matched college access to restricted-use data from the National Health Interview Surveys between 1984 and 2000. Because a respondent’s state of residence at the time of interview may not reflect their state of residence at the time of college choice, we assigned college access based on a respondent’s state of birth in order to avoid the potential that relocation into states with better college access (by the parents or children) confounds our results. Studying a range of health-related behaviours and health outcomes, we examined two sets of cohort-year-state fixed effects regression models, which all included state-level economic and policy covariates:
- The reduced-form effect of college access on college education, income, and health outcomes
- The effect of college education on income and health outcomes, with college access instrumenting for college education.
We present heat maps of the change in college access for two- and four-year colleges in US states between 1960 and 1996 in Figures 1 and 2. Changes in college access varied substantially by state, and the correlation between changes in two-year college access and in four-year college access was relatively low (we conducted a separate analysis that confirmed our measures of access were not highly correlated and did not predict the included age-17 state-level covariates).
Figure 1 Growth in public two-year colleges per capita, 1960-96
Source: Cowan and Tefft (2020).
Figure 2 Growth in public four-year colleges per capita, 1960-96
Source: Cowan and Tefft (2020).
The effects of college access and educational attainment on health
In order for college access to plausibly influence health, it would need to alter education outcomes. So we first estimated the effect of two- and four-year college access on years of schooling. Using data from the US Census Bureau, we found that public two-year college access significantly increased years of schooling for both non-Hispanic and Hispanic whites. Figure 3 plots the 1960 and 1996 state-level cohort differences in public two-year college access against average differences in years of schooling, along with the best fit regression line (the size of each bubble reflects state population).
Figure 3 Relationship between schooling and public two-year college access by state
Source: Cowan and Tefft (2020).
In the NHIS sample, the schooling effect was borne out only for non-Hispanic whites, and we suspect that the shorter timeframe for those data may be a factor (we discuss this more in our conclusion below). For that sub-population, we found that a one standard deviation increase in public two-year schools per capita resulted in a 1.2 percentage point decrease in the probability that a respondent is a current smoker, and an increase in exercise frequency of 0.2 events per week. There was no statistically significant effect on binge drinking or obesity. There was also a statistically significant increase (of 0.7 percentage points) in reporting excellent or very good health, a decrease (of 0.9 percentage points) in the probability of four or more visits to the doctor in the previous year, and decrease (of 1.4 percentage points) in the probability that the respondent has a cardiovascular condition. However, several other health outcomes were unaffected. For women, the results were more pronounced for doctor visits, smoking, and cardiovascular conditions, while for men, they were more pronounced for self-reported health and exercise frequency.
When we instrumented for years of schooling using college access in 2SLS regressions, we first estimated the effects of years of schooling on income to check consistency with previous studies and found that one more year of school increased income by about 14%. In 2SLS regressions with the previously discussed outcomes as dependent variables, we typically found somewhat more pronounced effects with the same coefficient signs (relative to OLS regression results), with the exception of doctor visits, where the sign flipped. Many of the first-stage F-statistics suggested that the instrument was weak, however, so we do not report these results in our paper.
Expanding two-year college access improves self-reported general health and has a beneficial effect on health-related behaviours, including exercise and smoking. Although our findings on the long-run health impact are limited to a reduction in the probability of having a cardiovascular condition, the mean age of respondents in our NHIS sample is approximately 36, so respondents may not yet have reached an age where the health impacts are evident. In future research, we plan to extend our analysis to an older sample in order to study longer-run effects. Future research might also include an investigation of the mechanisms through which health behaviours and subsequent health outcomes are affected – for example, through peer effects due to changed college peer groups or through improved health knowledge or investments in health.
Authors’ note: Our academic work on this subject benefited from funding from the Robert Wood Johnson Foundation Evidence for Action (E4A) programme (grant #76082).
Buckles, K, A Hagemann, O Malamud, M Morrill, and A Wozniak (2013), “The Effect of College Education on Health”, NBER Working Paper 19222.
Cowan, B W, and N Tefft (2020), “College Access and Adult Health”, NBER Working Paper 26685.
Currie, J, and E Moretti (2003), “Mother's education and the intergenerational transmission of human capital: Evidence from college openings”, The Quarterly Journal of Economics 118(4): 1495-1532.
Cutler, D M, and A Lleras-Muney (2010), "Understanding differences in health behaviors by education", Journal of Health Economics 29(1): 1-28.
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(5): 877-895.
Galama, T J, A Lleras-Muney, and H van Kippersluis (2018), “The Effect of Education on Health and Mortality: A review of experimental and quasi-experimental evidence”, NBER Working Paper 24225
Grimard, F, and D Parent (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 (2000), "The Human Capital Model", in Handbook of Health Economics, Vol. 1, Elsevier, pp. 347-408.
Grossman, M (2015), “The relationship between health and schooling: What’s new?”, NBER Working Paper 21609.
Kamhöfer, D, H Schmitz, and M Westphal (2018), “Heterogeneity in Marginal Non-Monetary Returns to Higher Education”, Journal of the European Economic Association 17(1): 205-244.
Oreopoulos, P, and U Petronijevic (2013), "Making College Worth It: A Review of the Returns to Higher Education", The Future of Children 23(1): 41-65.
1 http://money.cnn.com/2017/05/11/pf/college/tennessee-free-community-college/index.html (last accessed February 18, 2020).
2 https://www.ny.gov/programs/tuition-free-degree-program-excelsior-scholarship (last accessed February 18, 2020).
3 https://www.sacbee.com/news/politics-government/capitol-alert/article239574598.html (last accessed February 18, 2020).
4 For overviews of recent literature in this area, see Grossman (2015) and Galama et al. (2018).