Polarization of life expectancy across countries: Does biological and cultural distance to the health technological frontier matter?
| DOI | http://doi.org/10.1111/sjpe.12167 |
| Author | Wen‐Shuenn Deng,Yi‐Chen Lin,Ming‐Tien Tsai |
| Date | 01 July 2018 |
| Published date | 01 July 2018 |
POLARIZATION OF LIFE EXPECTANCY
ACROSS COUNTRIES: DOES BIOLOGI-
CAL AND CULTURAL DISTANCE TO
THE HEALTH TECHNOLOGICAL
FRONTIER MATTER?
Wen-Shuenn Deng*, Yi-Chen Lin* and Ming-Tien Tsai**
ABSTRACT
This study examines the evolution of the distribution of life expectancy (LE) at
birth and at ages 20 and 60 years using the stochastic kernel framework. It finds
that the period between 1990 and 2013 exhibits a decrease in LE at birth disper-
sion and also shows that r-divergence and polarization of LE at ages 20 and 60
occur across countries. Further deterioration (improvement) in relative position
of countries with low (high) starting LE is observed in some sections of the LE
distribution. The dispersion in LE and intradistribution movements can be attrib-
uted to genealogically transmitted human barriers to technology diffusion.
II
NTRODUCTION
What explains cross-country differences in health? Is disparity in life expec-
tancy (LE) rooted in long-term history? These questions are important
because LE has been found to affect a wide range of human behavior. An
emerging strand of the theoretical literature argues that divergence in intergen-
erationally transmitted biological and cultural traits between populations cre-
ates barriers to imitate and communicate, which in turn delays development
diffusion (Spolaore and Wacziarg, 2009, 2013, 2014, 2016). As a consequence,
biological and cultural relatedness to the health technological frontier plays
an important role in the distribution of development outcome across coun-
tries. Because of the regression framework used, the extant studies exclusively
look at the behavior of the conditional mean of LE and therefore are uninfor-
mative about why LE is unequal across countries.
In this study, we present evidence that highlights a different dimension: the
entire cross-sectional distribution of LE. Specifically, we ask whether genetic
distance to the world technological frontier, which has been used in previous
studies as a summary statistics for divergence in norms and values from the
*Tamkang University
**Academia Sinica
Scottish Journal of Political Economy, DOI: 10.1111/sjpe.12167, Vol. 65, No. 3, July 2018
©2018 Scottish Economic Society.
248
technological frontier, helps to explain the observed cross-country health
inequality. We use the nonparametric distribution dynamics approach of
Quah (1996, 1997), that is, stochastic kernel, and the bootstrap test for multi-
modality of Silverman (1981) to reveal the external shape dynamics and
intradistribution mobility of the cross-country distribution of life expectancy
(LE) among newborns, young adults, and the elderly across 174 countries
over the period from 1990 to 2013. We then examine whether normalizing LE
by the average of other countries in the same decile of genetic distance to the
world technological frontier (instead of the world average) significantly alters
the shape of the distribution.
1
If conditioning on the genetic distance to the
health technological frontier reduces the dispersion of the LE distribution
and/or removes the twin-peak feature from the distribution, then one can infer
that populations with similar biological and cultural differences to the techno-
logical frontier have similar LE. In other words, LE inequality would be much
smaller if countries have similar genealogical relatedness to the technological
frontier.
Our empirical results are threefold. First, while the dispersion of LE0 has
decreased, the variation in life expectancy at age 20 (LE20) and life expec-
tancy at age 60 (LE60) has increased sharply. Second, despite that the null
hypothesis of unimodality is not rejected in both 1990 and 2013 for LE0, for
life expectancy at age 20 (LE20), and life expectancy at age 60 (LE60), it is
rejected at the 10% level in 2013. The results together suggest that cross-coun-
try dispersion in LE decreased among newborns. Nevertheless, dispersion
increased and convergence clubs emerged in the distributions of LE among
young adults and the elderly. Third, after conditioning on the genetic distance
to the world technological frontier, dispersion and convergence clubs in LE0,
LE20, and LE60 significantly diminished. This implies that countries with sim-
ilar genetic distance to the world technological frontier exhibit similar LEs.
To the best of our knowledge, this is the first evidence whereby cross-country
differences in long-term genealogically transmitted human traits give rise to
health dispersion and polarization.
It is well understood that through gradual diffusion, technologies invented
in developed countries also raise human well-being in developing countries
(Eaton and Kortum, 1999). However, technologies developed elsewhere take
time to disseminate to the population.
2
Thus, a possible determinant of popu-
lation health is the absorptive capacity to internalize health technologies that
have been developed externally (Deaton, 2004; Cutler et al., 2006). What
explains the cross-country differential in the length of the process by which
new health institutions and technologies are absorbed into the population?
Technological differences are highly persistent over the very long run. In fact,
technology adoption in 1500 AD is a strong predictor of technology adoption
1
Quah (1997), Bandyopadhyay (2012), and Peron and Rey (2012) use conditioned stochas-
tic kernel to investigate whether countries/states that cluster into a convergence club in
income are trading partners or neighbors.
2
Bloom and Canning (2001) find that on average technological advances in health take 60
years to diffuse from developed to developing countries.
POLARIZATION OF LIFE EXPECTANCY 249
Scottish Journal of Political Economy
©2018 Scottish Economic Society
To continue reading
Request your trial