Do Education and Income Really Explain Inequalities in Health? Applying a Twin Design

• DOI: http://doi.org/10.1111/sjoe.12130
• Author: U.‐G. Gerdtham,C. H. Lyttkens,P. Nystedt,P. Lundborg
• Date: 01 January 2016
• Published date: 01 January 2016

Scand. J. of Economics 118(1), 25–48, 2016

DOI: 10.1111/sjoe.12130

Do Education and Income Really Explain

Inequalities in Health? Applying a Twin

Design∗

U.-G. Gerdtham

Lund University, SE-220 07 Lund, Sweden

ulf.gerdtham@nek.lu.se

P. Lundborg†

Lund University, SE-220 07 Lund, Sweden

petter.lundborg@nek.lu.se

C. H. Lyttkens

Lund University, SE-220 07 Lund, Sweden

carl_hampus.lyttkens@nek.lu.se

P. Nystedt

J¨

onk¨

oping University, SE-551 11 J¨

onk¨

oping, Sweden

paul.nystedt@ju.se

Abstract

We apply a twin design to examine the relationship between health and education and

income. The estimated associations between health and education and income, controlling for

unobserved endowments, at the twin-pair level, are lower than estimates obtained via ordinary

least-squares (OLS) on the same sample. Thus, OLS-based effects of education and income

are biased, exaggerating the contribution of education and income to health inequality. The

main part of health inequality is explained by within-twin-pair fixed effects, incorporating

family background and genetic inheritance. It appears that education and income policies

have less to offer for reducing health inequality than is usually assumed.

Keywords: Education; health inequality; income; twins

JEL classification:I10; I12; I14

†Affiliated with IZA Bonn.

∗We are grateful to Gustav Kjellsson, Rachel Knott, and two anonymous referees for

comments. Financial support from the Swedish Council for Working Life and Social Re-

search (FAS) (dnr 2012-0419) and the Swedish Research Council (dnr 2014-646) is gratefully

acknowledged. The Health Economics Program (HEP) at Lund University also receives core

funding from FAS (dnr. 2006-1660), the Government Grant for Clinical Research (ALF), and

Region Skane (Gerdtham).

C2015 The Authors. The Scandinavian Journal of Economics published by John Wiley & Sons Ltd on behalf of The

editors of The Scandinavian Journal of Economics.

This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License,

which permits use and distribution in any medium, provided the original work is properly cited,the use is non-commercial

and no modifications or adaptations are made.

26 Do education and income really explain inequalities in health?

I. Introduction

Numerous studies report a strong socioeconomic gradient in health and

longevity, regardless of the population studied and regardless of how so-

cioeconomic status and health are measured (e.g., Ettner, 1996; Smith,

1999; Bloom and Canning, 2000; Gerdtham and Johannesson, 2000, 2002,

2004; Benzeval and Judge, 2001; Deaton, 2003; van Doorslaer and Kool-

man, 2004; Baum and Ruhm, 2009). Despite improvements in average

health status over recent decades, this health gradient has persisted and

has even increased in most western countries (e.g., Mackenbach et al.,

2003; van Doorslaer and Koolman, 2004; Kunst et al., 2005; Shkolnikov

et al., 2011). Thus, reducing the magnitude of health inequality still poses

a challenge, and has become a major policy objective for many European

governments (Marmot et al., 2010).

Recently, the Marmot Review suggested that the strong association

between health and socioeconomic status implies that socioeconomic in-

equalities have to be reduced in order to reduce socioeconomic-related

inequalities in health (Marmot et al., 2010; Marmot, 2012). Accordingly,

policy measures affecting socioeconomic inequalities, such as income re-

distribution and publicly financed education, are taken to influence health

inequalities. This conclusion is based on the strong assumption that the as-

sociation between socioeconomic status and health reflects a causal effect

running from the former to the latter, which has attracted extensive debate

and disagreement in the literature (e.g., Smith, 1999; Deaton, 2002; Cutler

et al., 2008; Case and Paxson, 2011). This lack of consensus reflects the

incomplete nature of our knowledge about underlying causal mechanisms

and channels behind the disparities in health. Policies aimed at reducing

disparities might therefore easily be ineffective, inconsistent, and even coun-

terproductive (Deaton, 2011), and so there is an urgent need to improve

our understanding of the origins of socioeconomic health inequalities.

For policy purposes, the usefulness of any analysis of health and socioe-

conomic status critically hinges upon whether the estimated health function

reflects causal effects of socioeconomic factors on health. In practice, this

is tremendously difficult to prove (Deaton, 2011). There are two funda-

mental conceptual problems. First, causality might run from health status

to, for example, income, rather than the other way around. Second, there

might be some third, unobserved factor, perhaps of genetic or environmen-

tal origin, that determines both health and income (Fuchs, 1982; Smith,

1999; Mackenbach, 2003; Deaton, 2011). Similar causality issues hold for

education and other socioeconomic factors. One limitation of the major-

ity of prior studies is that they do not consider reverse causality or the

unobserved endowment and heritability that are correlated with socioeco-

nomic factors. Thus, the estimated contributions of socioeconomic factors

C2015 The Authors. The Scandinavian Journal of Economics published by John Wiley & Sons Ltd on behalf of The

editors of The Scandinavian Journal of Economics.