Health literacy and education: evidence from the English Longitudinal Study of Ageing

ABSTRACT

We estimate the causal effect of compulsory schooling on health literacy using data from the English Longitudinal Study of Ageing. We exploit the quasi-experimental setting produced by the UK’s 1944 Education Act. We estimate a positive effect of one additional year of schooling on health literacy among women, no significant effects among men. This result is in line with previous findings about the positive effects of compulsory schooling on own health among women and supports the idea that more schooling might have generated efficiency gains in the health production due to improved health literacy.

KEYWORDS:

• Health literacy
• compulsory schooling
• instrumental variables
• ELSA

JEL CLASSIFICATION:

• I20
• I20
• I21
• I28

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

¹ The health literacy test available in ELSA is more synthetic compared to ad hoc surveys indicators. However, it is correlated with higher mortality (Bostock and Steptoe 2012) and low participation to cancer screening programmes (Kobayashi, Wardle, and von Wagner 2014) as other more comprehensive measures are.

² Among eligible individuals, 96% were assessed as able to do the test. Sight difficulties is one of the conditions – reported by interviewers more frequently – that impedes doing the test.

³ Following the empirical strategy described in Section V, we report in Table A.1 estimates showing the positive effects of the reform on limitations in activities (instrumental or not) of daily living among women for our working sample.

⁴ The vast majority of individuals, belonging to the cohorts we are interested in, took the health literacy test for the first time in wave 2. In our sample, 85% of individuals took the test in wave 2; 15% in wave 5 for the first time.

⁵ We computed also bootstrapped SEs, which are available upon request. Although larger, they lead to the same conclusions.

⁶ Focusing on narrower windows around the cut-off or conditioning for higher order polynomials of date of birth helps controlling also for cohort patterns of age-specific mortality rates, as discussed in Banks and Mazzonna (2012). Furthermore on this regard, we are likely to estimate a lower bound of the effect, since we expect higher mortality among older cohorts on the left side of the cut-off, assuming there is a positive selection on survival. Finally, there is additional evidence on the insignificant direct effect of the reform on mortality in Clark and Royer (2013).

⁷ The health literacy test explicitly does not measure memory. Before starting the test, the interviewer clarifies the following: You do not have to memorize the card, as you will be able to look at it while answering the questions.

⁸ See discussion in Angrist and Pischke (2009b).

⁹ The economics literature on the mechanisms through which education affects health is broad; as discussed in Clark and Royer (2013), many channels have been identified especially related to the income–health relationship (see for instance Cutler and Lleras-Muney 2006, Evans & Moore 2011, 2012), we focus here on the empirical evidence specifically related to the reform in question.

¹⁰ This applies especially to the old cohorts we analyse.

¹¹ The home versus market sector distinction borrows from Becker (1981), when discussing spouses’ specialization (Pollak 2012). Activities belonging to the home sector – e.g. childcare or grocery shopping – are typically relatively more health related than those of the market sector.