Upload
others
View
5
Download
0
Embed Size (px)
Citation preview
Height, Health and Cognitive Function at Older Ages: Cross-
National Evidence from Europe
Cahit Guven
Deakin University
Wang-Sheng Lee
RMIT University
March 23, 2011
Preliminary. Please do not cite or quote.
Abstract
Previous research has found that height is correlated with cognitive functioning at older ages.
It therefore makes sense to ask a related question: are people from countries where the
average person is relatively tall smarter? Using data from the Survey of Health, Ageing, and
Retirement in Europe (SHARE), we find empirical evidence that this is the case, even after
controlling for self-reported childhood health, self-reported childhood abilities, parental
characteristics and education. We find that people from countries with relatively tall people,
such as Denmark and the Netherlands, have on average superior cognitive abilities compared
to people from countries with relatively shorter people, such as Italy and Spain.
Keywords: Height, Self-reported health, Cognitive function
JEL Classification: I10
This paper uses data from SHARELIFE release 1, as of November 24th 2010 or SHARE release 2.3.1, as of July
29th 2010. The SHARE data collection has been primarily funded by the European Commission through the 5th
framework programme (project QLK6-CT-2001- 00360 in the thematic programme Quality of Life), through
the 6th framework programme (projects SHARE-I3, RII-CT- 2006-062193, COMPARE, CIT5-CT-2005-
028857, and SHARELIFE, CIT4-CT-2006-028812) and through the 7th framework programme (SHARE-
PREP, 211909 and SHARE-LEAP, 227822). Additional funding from the U.S. National Institute on Aging
(U01 AG09740-13S2, P01 AG005842, P01 AG08291, P30 AG12815, Y1-AG-4553-01 and OGHA 04-064,
IAG BSR06-11, R21 AG025169) as well as from various national sources is gratefully acknowledged (see
www.share-project.org/t3/share/index.php for a full list of funding institutions).
1
1. Introduction
Previous research on the influence of early-life conditions on cognitive development
suggests that socioeconomic conditions in childhood and early life experiences have
important influences on cognitive development and abilities in childhood and adolescence, as
well as in young and middle adulthood. For example, children from poor backgrounds show
worse verbal and achievement outcomes in the first 5 years of life (Duncan et al. 1994). Low
socioeconomic status in childhood has also been associated with cognitive function in middle
age, net of years of education completed (Kaplan et al., 2001). It is only in recent years that
researchers adopting a life course approach have begun to trace the origin of cognitive
functioning in old age to early life conditions.1 For example, using longitudinal data in
conjunction with retrospectively collected childhood data, Everson-Rose et al. (2003), Wilson
et al. (2005) and Zhang et al. (2008) all find that higher socioeconomic status during
childhood are weakly associated with a higher absolute level of cognitive function in old age.
More recently, van den Berg et al. (2010) use a unique Dutch longitudinal dataset to examine
the role of early life socio-economic circumstances in protecting individuals from cognitive
decline in the face of adverse events later in life. They show that the cognitive abilities of
those who suffer from strokes later in life are more heavily affected if individuals were born
in adverse socioeconomic conditions.
In the absence of reliable data from early childhood, several recent studies use adult
height as a marker of childhood circumstances.2 For example, Abbot et al. (1998) examine
the relationship of height to later life cognitive function in a sample of elderly Japanese-
American men and find the prevalence of poor cognitive performance declined consistently
with increasing height. Their findings suggest that efforts to improve prenatal and early life
conditions to maximize growth in childhood and adolescence could diminish or delay the
expression of cognitive impairments that occur later in life. Case and Paxon (2008a) further
suggest that height could be an indicator of higher cognitive potential in the sense that people
who do not reach their full genetic height potential do not reach their full genetic cognitive
potential either. They provide evidence that taller individuals are more likely to earn more,
not because of their heights per se, but because of the cognitive skills with which height is
correlated. Using data from the U.S. Health and Retirement Study (HRS), Case and Paxson
1 See Factor-Litvak and Susser (2004) for a recent overview.
2There is a large literature in economic history that uses height as a key measure of physical welfare and the
standard of living. See, for example, the surveys by Steckel (1995, 2009). Aside from genetics, it has been
established that height is influenced by childhood nutrition and disease (e.g., Fogel, 1993; Peck and Lundberg,
1995). Hence, height is commonly seen as a useful marker of overall childhood conditions.
2
(2008b) document a strong association between self-reported height and cognitive function in
later life. Similarly, Maurer (2010) complements the evidence presented in Case and Paxson
(2008b) by examining the later-life cognition of seniors in Latin America and the Caribbean
using data from the Survey on Health, Well-being and Aging in Latin America and the
Caribbean, 2000 (SABE). He finds that that height displays a strong positive association with
later life cognition, which seems somewhat larger for women than for men. These studies
build on the earlier work of psychologists who have previously noted that height appears to
be positively correlated with intelligence (e.g., Humphreys et al, 1985; Jensen and Sinha,
1993; Johnson, 1991; Tanner, 1969).
If within country studies such as those by Case and Paxson (2008b) and Maurer
(2010) find a significant association between height and cognitive function in later life, it is
natural to also ask whether such a correlation exists for individuals across countries. Do
people from countries where the average native person is relatively tall have superior
cognitive abilities compared to people from countries where the average native person is
relatively shorter? It is well known that the average heights differ across nationalities
considerably. Do the taller Austrians and Danes, for example, have higher cognitive abilities
than the shorter Italians and Spaniards?
In this paper, using data from the Survey of Health, Ageing, and Retirement in Europe
(SHARE), we examine the relationship between height and cognitive function in a sample of
older Europeans. Our objectives are twofold. First, we add to the current literature by
extending the work of Case and Paxon (2008b) and Maurer (2010) by examining the
relationship between height and later life cognition in 15 additional European countries not
previously analyzed. Second, we extend the within-country analysis to a cross-country
analysis to determine if countries with relatively taller people have higher levels of cognitive
function than countries with relatively shorter people.
2. Background
2.1 Cross-Country Variation in Height
Garcia and Quintana-Domeque (2007) document the evolution of adult heights in
Europe in the period 1950-1980. They find that average height in the Northern European
countries (Austria, Belgium, Denmark, Finland, Ireland, and Sweden) is higher than in the
Southern ones (Greece, Italy, Portugal, and Spain) for both males and females. Hatton and
Bray (2010) extend the database constructed by Garcia and Quintana-Domeque (2007) by
going through a variety of historical records to include the average heights of men by birth
3
cohorts from 1856–60 to 1976–80. Eveleth and Tanner (1976, 1990) produced a world-wide
overview of variations in growth among children aged 2 to 16 years. They used data from
studies undertaken from the 1950s to the 1980s that were based on nationally representative
samples or on samples from large cities within the countries studied. Identical growth curves
were observed for European countries. However, at age 16, they also found that children in
northern European countries were on average taller than children in southern European
countries. Similarly, when De Groot et al. (1991) compared the height of elderly subjects
born between 1913 and 1918 in 19 cities across Europe, they found that these subjects were
tallest in northern European populations.3 In this paper, we exploit this variation in height
across European countries and explore in more detail whether adult height is useful as a
marker of cognitive ability.
2.2 Possible Explanations for Cross-Country Variation in Cognitive Ability
Technological progress of nations, a key ingredient of a country‟s economic success
and the wealth and well-being of its citizens, has been shown to be related to average national
levels of intelligence (e.g. Gelade, 2008). It is therefore of great interest how cross-country
differences in cognitive abilities arise. Several hypotheses have been put forth in attempts to
explain the variation in the global distribution of cognitive ability. These include exposure to
education and other cognitively challenging environments such as non-agricultural labor,
differing levels on inbreeding across countries, the effects of temperature and climate, as well
as the effects of variation in the intensity of infectious diseases.
Using data on intelligence quotient (IQ) scores for 81 nations and focusing on
bivariate correlations, Barber (2005) finds that average national IQ to be correlated with
enrolment in secondary school (r = 0.72), illiteracy (r = -0.71), agricultural labour (r = -0.70)
and gross national product (r = 0.54). He also proposed that health and nutrition may affect
intelligence, and found that average national IQ correlated negatively with rates of low birth
weight (r = -0.48) and with infant mortality (r = -0.34).
3 Although there is a large genetic component to heights within populations, the contribution of genetics to
variation in mean heights across populations is much smaller. For example, Beard and Blaser (2002) argue that
the marked increase in heights observed throughout the developed world during the twentieth century occurred
too rapidly to be due to selection and genetic variation. Environmental factors are likely to be important. For
example, Cavelaars et al. (2000) found high and significant correlations between average height and infant
mortality for each of the 5-year birth cohorts born between 1920 and 1970 (their Pearson correlation coefficients
averaged 0.8 for men and 0.6 for women), suggesting that infant mortality and height may indeed have common
determinants.
4
Sadat (2008) and Woodley (2009) explore the hypothesis that inbreeding and the
associated reduced phenotypic quality is a cause of the variation in cognitive ability across
the world. Jensen (1983) finds that the effect of inbreeding on the intelligence of the offspring
of first cousins amounts to about 5 IQ points. Consanguineous marriages (i.e., a marriage
between first or second cousins) account for a significant percentage of marriages in some
countries. Although stigmatized in the West, such marriages are common in many Middle
Eastern countries such as Saudi Arabia (39.7%) and Qatar (44.5%), as well as African
countries such as Sudan (50.1%) and Nigeria (51.2%).4 In support of this hypothesis, Sadat
(2008) and Woodley (2009) found significant cross-national correlations in the range of 0.6
to 0.8 between average IQ and measures of inbreeding.
Lynn (1991) and Rushton (1995) proposed that temperature and climate provide
important Darwinian selective pressures for intelligence, with cold climates selecting for
higher intelligence, because low temperatures provide more fitness-related problems for
humans that must be solved through cognitively demanding means, and through more
complex social organization. Some empirical support for this hypothesis was reported in
Templer and Arikawa (2006) who found that persons in colder climates tend to have higher
IQ scores.
Finally, Eppig et al. (2010) have recently provided empirical evidence using a sample
of over 100 countries that average national intelligence correlates significantly and negatively
with rates of infectious disease. This is possible because parasitic infection may intermittently
cause the redirection of energy away from brain development during the crucial years of
childhood development.
2.3 Do Cross-Country Differences in Height have any Economic Significance?
In this paper, we explore yet another hypothesis to explain cross-country variation in
cognitive abilities – the role of height. This is a natural extension of the significant links
found by Case and Paxon (2008b) and Maurer (2010) between height and cognitive function
in later life using within-country studies.
Somewhat related to this hypothesis are papers by Angus Deaton and his co-authors
exploring the significance of cross-country differences in height. Deaton (2007) analyses the
link between adult height, disease and national income using data on 43 countries from the
Demographic and Health Surveys. With the exception of Africa, he finds there is a general
4 These figures are taken from Appendix A in Woodley (2009).
5
interregional correspondence between height and national income. Over time, as real incomes
have grown, heights have grown too.
In a related paper, Bozzoli, Deaton and Quintana-Domeque (2009) focus their
analysis on adult height and childhood disease in the US, England and ten European countries
where more detailed household level data are available. They find that both within and
between these countries, there is a close relationship between income per capita and height.
Their findings also suggest that the direction of causality does not appear to run from income
to height, as they find that the disease environment in infancy is the most important
determinant of adult height, not the level of income per head. Indirectly, the role of height is
also possibly related to the Eppig et al. (2010) hypothesis involving the role of infectious
diseases in influencing cognitive functioning, as children who get sick when they are very
young might suffer some physical developmental consequences.5
2.4 Possible Links between Height and Cognitive Ability
To date, the precise mechanisms underlying the relationship between height and IQ
are still not well understood. Case and Paxon‟s (2008b) study highlights the crucial role of
education as a potential pathway linking height and cognitive function in later life. They find
that there is a statistically significant positive association between cognitive function and self-
reported height, which declines considerably once they control for education. They also
highlight a positive association between education and height. Taken together, these findings
suggest that early-life conditions have an effect on later-life cognition results partly due to
higher levels of schooling among children with higher socioeconomic status, which may, in
turn, protect cognitive function at older ages.
Other mechanisms that produce a positive correlation between height and intelligence
have been suggested. It is possible that an unmeasured factor simultaneously affects cognitive
ability and height. For example, Lynn (1990) emphasizes the role of nutrition, arguing that
the most straightforward explanation of the positive association between height and
intelligence is that both are functions of nutrition. He argues that improvements in nutrition
have led to parallel increases in height, head circumference, brain size, and to improved
neurological development and functioning of the brain. On the other hand, biological factors
could be important. Prokosch, Yeo and Miller (2005) suggest that it occurs via an alternative
5 They could, however, be very distinct hypotheses. As we report in Section 4, to the extent that self-reported
information on childhood diseases is reliable, we find that childhood diseases have little effects on reducing the
statistical significance of height in our cognitive function regressions we estimate in Table 7.
6
mechanism – people with greater genetic quality and developmental health may
simultaneously have higher intelligence and greater stature. Berger (2001) highlights the
existence of insulin-like growth factors that affect body growth while also influencing areas
of the brain in which cognition occurs.
Twin studies have been used to identify the roles played by a common environment
and shared genes. Sundet et al. (2005), using conscription data of Norwegian twins, conclude
that the environment plays a large role and is responsible for 65 percent of the height-
intelligence correlation, with genes responsible for 35 percent of the observed correlation.
Beauchamp et al. (2010), using a sample of Swedish twins, find results that are very similar
to those reported by Sundet et al. (2005). Silventoinen et al. (2006), however, found in
several samples of Dutch twins that the association between height and intelligence is
primarily genetic in origin.
The role played by nutrition in linking height and cognitive functioning has also been
supported by twin studies. Black et al. (2007) find that, on average, the twin born at the
higher birth weight is significantly taller in adulthood and scores significantly higher on IQ
tests. Similarly, using data from the Minnesota Twin Registry, Behrman and Rosenzweig
(2004) find fetal growth (birth weight divided by gestation) to be significantly associated
with height and years of completed schooling in adulthood.
Kanazawa and Reyniers (2009) propose a somewhat more lengthy explanation
comprising of three separate mechanisms and involving genetic evolution over time. The first
mechanism involves the assortative mating of tall men and beautiful women. As height is
desirable in men and physical attractiveness is desirable in women, there should be
assortative mating between tall men and beautiful women (and short men and less beautiful
women). Since both height and physical attractiveness are heritable, this will create a
correlation among their children between height and physical attractiveness, where tall people
(both men and women) are more beautiful than short people. The second mechanism involves
the assortative mating of intelligent men and beautiful women. As intelligent men tend to
attain higher status, at least in the evolutionarily novel environment in recent history, and
high status is desirable in men, and because physical attractiveness is desirable in women,
there should be assortative mating between intelligent (and thus high-status) men and
beautiful women. Since both intelligence and physical attractiveness are heritable, this will
create a correlation among their children between intelligence and physical attractiveness,
where more attractive people are more intelligent than less attractive people. Finally, the
correlation between height and physical attractiveness (produced by the first mechanism
7
above) and correlation between intelligence and physical attractiveness (produced by the
second mechanism above) will create a second-order correlation between height and
intelligence.
3. Data
In this study, we use data on health and cognitive functioning from Waves 1 and 2 of
SHARE (Release 2.3.1 of July 29, 2010). The first wave was fielded in 12 countries in
2004/2005: Austria, Belgium, Denmark, France, Greece, Germany, Italy, the Netherlands,
Sweden, Switzerland, Spain and – one year later – Israel. In all countries, probability samples
of nationally representative samples of the community-based population aged 50 and older
were drawn. The 31,000 interviews conducted in that period correspond to a weighted
average household response rate of 61 percent, ranging from 39 percent in Belgium and
Switzerland to 79 percent in France (a thorough description of methodological issues is
contained in Börsch-Supan and Jürges, 2005). The second wave in 2006/2007 opened the
longitudinal dimension, but also collected baseline data from three further countries: the
Czech Republic, Poland and – after some delay – Ireland.
We also use data from the third wave of data collection for SHARE. This data, which
was in the field from October 2008 to May 2009, focuses on people's life histories and is
otherwise commonly referred to as SHARELIFE. Waves 1 and 2 of SHARE provide little
information about what happened earlier in the lives of survey respondents. SHARELIFE
gathered more detailed information on important areas of our respondents‟ lives, ranging
from partners and children over housing and work history to detailed questions on health and
health care. It therefore complements the SHARE panel data by providing life history
information and enhancing our ability to understand how early life experiences and events
throughout life influenced the circumstances of the survey respondents. SHARELIFE data are
available for all the countries in SHARE with the exception of Ireland and Israel.
SHARE is designed to be cross-nationally comparable and is harmonized with the
U.S. Health and Retirement Study (HRS) and the English Longitudinal Study of Ageing
(ELSA). International comparability is achieved by ex-ante harmonization of the survey
instrument and all fieldwork procedures. The common questionnaire and interview mode, the
effort devoted to translation of the questionnaire into the national languages of each country,
and the standardization of fieldwork procedures and interviewing protocols are the most
important design tools adopted to ensure cross-country comparability (Börsch-Supan and
Jürges, 2005).
8
In SHARE, cognitive ability is measured using simple tests of orientation in time,
memory (registration and recall of a list of ten words), verbal fluency (a test of executive
function) and numeracy (arithmetical calculations). Participants are also asked to rate
subjectively their reading and writing skills. These tests are administered to all respondents
and are carried out after the first four modules (Cover Screen, Demographics and Networks,
Physical Health, and Behavioral Risks) of the questionnaire. The tests are comparable with
similar tests implemented in the HRS and ELSA, and follow a protocol aimed at minimizing
the potential influences of the interviewer and the interview process.6
The test of orientation in time consists of four questions about the interview date (day,
month, year) and day of the week. Unfortunately, this test shows very little variability across
respondents, with a majority of respondents answering all four questions correctly.
Nevertheless, we include it in this paper for comparability purposes with Case and Paxon
(2008b).
The test of memory consists of verbal registration and recall of a list of 10 words
(butter, arm, letter, queen, ticket, grass, corner, stone, book and stick). The respondent hears
the complete list only once and the test is carried out two times, immediately after the words
are read out (immediate recall) and at the end of the cognitive function module (delayed
recall). The raw total scores of both tests correspond to the number of words that the
respondent recalls.
The test of verbal fluency consists of counting how many distinct elements from a
particular category the respondent can name in a specific time interval. The specific category
used in SHARE is members of the animal kingdom (real or mythical, except repetitions or
proper nouns) and the time interval is one minute for all respondents.
The test of numeracy consists of a few questions involving simple arithmetical
calculations based on real life situations. Respondents who correctly answer the first question
are asked a more difficult one, while those who make a mistake are asked an easier one. The
last question is about compound interest, testing basic financial literacy. The resulting raw
total score ranges from 0 to 4. Finally, respondents are also asked to rate their reading and
writing skills on a five point scale: Excellent = 1, Very good = 2, Good = 3, Fair = 4, Poor =
5.
6 An important drawback of SHARE is that the exact same tests were administered to all respondents of the
same household and to the same individual over time. Repeated exposure to the same tests may induce learning
effects which are likely to improve the cognitive scores of some respondents.
9
Our height measure in SHARE is based on self-reported height (“How tall are you?”).
Although many studies have observed a very high correlation between measured height and
self-reported height (e.g., Palta, Prineas, Berman et al., 1982; Steward, 1982; Steward et al.,
1987; Bostrom and Diderichsen, 1997), these studies also found that using self-reported
height leads to a slight overestimation of the average height of the study population.
Moreover, some studies showed that this overestimation was larger among men, among older
age groups, and among lower socio-economic groups (Palta et al., 1982; Steward et al., 1987;
Bostrom and Diderichsen, 1997). However, where the focus is on cross-national comparisons
the main results will only be biased when this over- or underestimation also varies between
countries. Supporting the use of our data is that the large variations in average height we
observed between northern and southern European countries were also reported in studies in
which height was measured (Eveleth and Tanner 1976, 1990; de Groot et al. 1991). Further
reducing the need for a concern that the use of self-reported height in SHARE leads to bias is
that when we regressed measures of cognitive function on two available height measures in
ELSA (both self-reported height and nurse measured height), we found very similar results.
The measure of health we use in our analysis in SHARE is also based on self-reports.
Respondents are asked to self-assess their health on a five-point scale in SHARE. This survey
question has been widely used in health surveys and is meant to reflect overall health status.
In Wave 1, SHARE used the US version of this self-reported health scale (excellent, very
good, good, fair and poor) for some respondents, and also included the European scale (very
good, good, fair, bad and very bad) for other respondents. Respondents were given either the
US or the European scale randomly but not asked to answer both. In Wave 2, SHARE only
used the US scale for all respondents. From an analysis of the distribution of self-reported
health from both scales in Wave 1, it appears that responses are partly based on the specific
words in the response options. Differences in the mean answer (on a scale of 1 to 5) were
different for each country, suggesting that there is not an easy mapping between the scales.
Following Meenakshi et al. (2008), one possible way around this problem is to construct a
binary measure of self-reported health that makes the European and US scale responses as
comparable as possible: Those who report excellent, very good or good health on the
American scale are considered to be in “good” health, whereas those who report to be in fair
or poor health are classified as being in “bad” health. Using the European scale, those in very
good or good health are classified as being in “good” health while those reporting fair, bad or
very bad health are considered to be in “bad” health.
10
We use two measures of well-being. First, there is a question asking: “Have you ever
experienced depression in your life?” We code the binary response whereby „1‟ indicates that
they have ever been depressed, and „0‟ indicates that they‟ve never experienced depression
before. Second, levels of depression among participants in the SHARE survey were also
measured using the Euro-D 12-item scale, as this has been validated in earlier cross-European
studies of depression prevalence (Prince et al. 1999a, Prince et al. 1999b). Respondents were
asked questions with reference to presence of feelings of depression, pessimism, wishing
death, guilt, irritability, tearfulness, fatigue, sleeping troubles, loss of interest, loss of appetite,
reduction in concentration, and loss of enjoyment over the last month. In order to get a
measure of subjective well-being, we „reverse coded‟ the answers of respondents with 1 =
„absence of feeling' or 0 = „presence of feeling'. The sum can thus range between 0 and 12,
with a higher sum indicative of lower degrees of depression, and lower values corresponding
to more depression.
4. Results for Individual European Countries
In this section, we focus on replicating the country studies of Case and Paxson
(2008b) and Maurer (2010) who report significant associations between height and cognitive
function in later life. It is interesting to see if their results hold for many of the developed
European countries. Table 1 provides descriptive statistics of height, the seven cognitive
function variables we use in our analysis, as well as the three health variables and key control
variables.
[Table 1 about here]
Using pooled data, the estimated results in which these seven cognitive measures are
regressed on respondents‟ heights (measured in centimetres), age, race, gender country
dummies and survey wave are reported in Table 2. Focusing on the coefficient for height, for
all 15 European countries as a whole, it can be seen that it is highly statistically significant in
columns one to seven, suggesting that taller persons perform better on average in the
cognitive tests as compared to their shorter counterparts. For example, a person that is 10 cm
taller is expected to score 0.16 more points on the immediate recall test.
[Table 2 about here]
11
We next turn to examining the country specific results not reported in Table 2, which
is essentially a replication of the analysis by Case and Paxon (2008b) using the HRS for each
of the 15 European countries in the SHARE data set. We can see that the coefficient on
height is statistically significant in the majority of cases (Table 3). The main exceptions are
for Israel, where in six out of the seven cognitive measures, the height coefficient is not
significant, and for Ireland, where in four of the seven cognitive measures, the height
coefficient is not significant. The height coefficient is generally not significant for the test on
orientation in time (column 3) because of little variation in the test score, with most people
scoring close to full marks in the test (mean of 3.81 out of 4).
[Table 3 about here]
Table 4 examines the determinants of height, education and occupational choice. The
estimates presented here are based on 13 countries as both Ireland and Israel have no
childhood data collected via SHARELIFE available. The first column presents the result of a
regression of height on childhood correlates of height – father‟s occupation at age 14 (as a
proxy for childhood socioeconomic status) and self-reported childhood health status.
Respondents whose fathers were in white collar occupations when they were 14 years old are
on average 1.2 centimetres taller than those whose fathers were not, whereas respondents
who reported having “good” childhood health are likely to be 0.13 centimetres taller on
average. Looking at the determinants of years of schooling (column 2 of Table 4) and white
collar occupation (columns 3-4 of Table 4), our findings echo those reported in Case and
Paxon (2008b) – even controlling for father‟s occupation and childhood health, height
remains a significant determinant of education and occupation.
[Table 4 about here]
Table 5 presents the results of regressions similar to those in Table 2, with the
difference that extra control variables for education, father‟s occupation at age 14, own
occupation and childhood health have been added. Strikingly, despite the inclusion of
education as a control variable, a factor which largely diminished the role of height in similar
regressions estimated in Case and Paxon (2008b), the height coefficient is still statistically
significantly correlated with our seven measures of cognitive functioning and subjective
12
health. This suggests that in the SHARE data, height does not only operate via education in
affecting cognitive outcomes.
[Table 5 about here]
As a marker of childhood conditions, the residual effects of childhood health may
continue to be reflected and included in the coefficient on height in regressions on adult
outcomes if information on childhood circumstance is not accurately captured. Hence, in
addition to using self-reported childhood health status in Table 4, we consider additional
childhood history variables available in the SHARE data set. These relate to the home
environment in which the child was brought up as well as self-reported childhood abilities at
age 10. In column 1 of Table 6, the results of the height regression show that the signs on the
childhood history variables are as expected. For example, having parents that drank heavily
or having parents with mental problems led to poorer parenting and as a result poorer
childhood growth. As an indicator of household wealth in childhood, it can be seen that larger
dwellings are associated with better childhood growth, likely through the better provisions of
childhood resources. On the other hand, competition for such limited resources with other
family members is likely to lead to reduced growth, as reflected in the significant negative
coefficient in the variable „dwelling with more than 4 people‟.
Self-reported abilities in mathematics and language are significantly positively
associated with years of schooling and having a white collar occupation. Including these
childhood history variables have little effect, however, on reducing the significance of the
height coefficient in the regressions using education and occupation as the dependent variable
(columns 2-4 of Table 6).
[Table 6 about here]
In Table 7, we add the childhood history variables to the regressions on cognitive
function and health estimated in Table 5. Interestingly, comparing the coefficients for height
in Tables 5 and 7, we can see that the inclusion of more detailed childhood history variables
hardly has any effects on the size of the height coefficient for all seven cognitive outcomes.
[Table 7 about here]
13
Our results for 15 additional European countries provide more empirical support for
the link between height and cognitive functioning. However, it is less clear what channels
height operates through to affect cognitive functioning in later ages.
Our evidence suggests that education is not the main pathway that height affects
cognitive functioning, as including years of schooling did not affect the significance of the
height coefficient. We also tried including information on the highest degree earned, but this
made no substantive difference. It is likely that height possibly captures some other aspects of
early childhood health experiences that we are unable to measure, or alternatively some other
factors associated with cognitive functioning that we have not included in our models.
The adverse early-life conditions that affect cognitive functioning that have been
studied are mostly nutritional (e.g., Lynn, 1989; Kretchmer et al. 1996). However, exposure
to high levels of stress or illness during the critical childhood years – a much less researched
area – could also have important effects. In an attempt to test whether childhood health
shocks might have affected both height and cognitive development, we also tried including
detailed information regarding childhood health conditions in SHARE. These include:
whether or not missed school due to health problems, infectious diseases, broken
bones/fractures, asthma, other allergies, other respiratory problems, chronic ear problems,
severe headaches and migraines, epilepsy/seizures/fits, emotional/psychological problems,
appendicitis, diabetes/high blood sugar, heart trouble, leukaemia/lymphoma,
cancer/malignant tumor, diseases of the blood, diseases of digestive system, and upper
respiratory organs diseases. Despite restricting the first occurrences of these health conditions
to ages 0 to 5, we found that these variables did not reduce the statistical significance for
height.
5. Cross-Country Differences in Cognitive Function
In this section, we extend the within-country analysis conducted thus far to a cross-
country framework. If tall people within each country demonstrate superior cognitive abilities
relative to shorter people, and this finding appears to be very robust for many different
countries, then it is natural to wonder if people from countries with relatively tall people are
smarter than people from countries with relatively shorter people.
5.1 Cross-Country Results using SHARE
Figure 1 provides average height by country for the both men and women, as well as
separately by gender. It can be seen that the Dutch are the tallest in the sample, with male
14
average heights of about 1.78 metres and female average heights of about 1.66 metres. The
Danes and the Swedes are also relatively tall, with average male heights over 1.77 metres and
average female height over 1.64 metres. On the other hand, people from Spain, Ireland and
Italy are the shortest in our sample. Our results largely correspond with the historical
European height data that is reported in Garcia and Quintana-Domeque (2007) and Hatton
and Bray (2010).
[Figure 1 about here]
Figures 2 to 6 provide country averages of the five cognitive measures in order to
highlight the raw differences across countries. There is considerable cross-country variation
in most measures, such as numeracy (Figure 2), verbal fluency (Figure 4), immediate recall
(Figure 5) and delayed recall (Figure 6). The sole exception is the average date score. But that
lack of variation is easily explained – with most people scoring close to full marks for the
test, there is simply not much room for cross-country variation. A casual glance at the figures
suggests that height could be correlated with cognitive functioning, as countries with the
shortest people – Italy and Spain – also tend to perform the most poorly on all cognitive
measures. On the other hand, the Northern European countries with relatively taller people
tend to score better on each test.
[Figures 2-6 about here]
In order to examine the cross-country relationship between height and each of the
cognitive measures, we first regressed height and each of the cognitive and health outcomes
on a full set of person-level covariates to control for people‟s different characteristics (see the
list of covariates in Table 6 and 7) and country dummies. Table 8 provides the summary
results of the 14x11 = 144 separately estimated regressions, with Spain as the omitted
reference country.
The correlations of the variables in Table 8 are presented in Table 9. We emphasize
that these cross-country correlations are based on the country dummies reflect adjustments
made for differences in individual level characteristics, and are not simply cross-country
correlations of country averages. The latter is the approach most commonly taken when only
aggregate country level data are available (e.g., in the empirical economic growth literature).
15
Our approach of building up country level aggregates from micro-level data helps us to better
adjust for the different characteristics that people from different countries have.
[Table 8 about here]
[Table 9 about here]
Our findings are striking. We had previously seen that there are strong positive
associations between height and cognitive function in each of the 15 countries (Table 3).
Here, we also find that countries with taller people have higher levels of cognitive function in
our cross-country comparisons. High correlations can be found, in particular, for verbal
fluency, immediate recall, and numeracy. While we report simple Pearson correlation
coefficients in the first row of Table 9, it is arguable that more appropriate correlation
measures rely only on the ordinality of the measures. We therefore also perform both
Kendall‟s and Spearman‟s rank correlation tests. Kendall‟s tau statistic is particularly suitable
for smaller data sets such as ours. Two-sided tests of the null hypothesis of no correlation
between the country dummies suggest that at the five percent level, there are significant
correlations between height and verbal fluency, height and immediate word recall, height and
delayed word recall, height and numeracy, and height and psychological well-being.
Plots of the cross-country relationships between height and the various measures of
cognitive function and health are given in Figure 7.
[Figure 7 about here]
5.2 Cross-Country Correlations of Height and Alternative Cognitive Measures
In this section, we provide estimates of the cross-country correlations of height with
some other commonly used national indicators of cognitive function. In addition to the results
we have already discussed with regards to the SHARE data, these additional results are
suggestive of the potential important information that variation in country-level average
heights contain.
Table 10 provides the correlation of the country dummies for height from column 1 of
Table 8 with several publicly available measures of cognitive function. First, we correlate
16
height with scores from the Programme for International Student Assessment (PISA).7 PISA
scores are generally regarded as informative and reliable indicators of cross-country
comparisons for cognitive ability (Hanushek and Woessman, 2008). In the top two rows of
Table 10, we can see that national height based on the SHARE data is highly correlated with
the performance of 15 year olds in the PISA mathematics score. However, it appears to not be
correlated with the PISA literacy score. The correlation between height and the PISA
mathematics score is significant at the five percent level according to Spearman‟s rank
correlation test and is nearly significant at the ten percent level according to Kendall‟s rank
correlation test.
Next, we also correlate height with various indices of talent and creativity taken from
Florida and Tinagli (2004). We can see in the lower half of Table 10 that national height is
highly correlated with all four measures of talent and creativity. The Pearson correlation
between height and the creativity index is 0.82, suggesting that countries with relatively tall
people are more innovative and creative, which could have long term implications for
economic growth. Hypothesis tests using both Kendall‟s and Spearman‟s rank correlation
tests find that height is significantly correlated with the creativity index, the tolerance index
and the technology index.
[Table 10 about here]
Clearly, the height measure we use does not correspond directly with measures for 15
year olds or the labor force as a whole. However, beyond what is given in Garcia and
Quintana-Domeque (2007) and Hatton and Bray (2010), height data by country and age
group does not appear to be readily available. Assuming the rank ordering of countries based
on height remains the same when other age groups are used, the results presented in Table 10
might be viewed as being suggestive of the potential uses that country-level height data might
provide.8
Previous work has focused on identifying important drivers of various national
indicators of policy importance, such as economic development (e.g., Deaton, 2007;
7 PISA (http://www.pisa.oecd.org) is a regular survey of 15-year olds which assesses aspects of their
preparedness for adult life. Four assessments have so far been carried out, in 2000, 2003, 2006 and 2009. 8 In spirit with Zipf‟s law, the rank ordering of countries based on height could provide useful information that
makes the actual collection of height data less important, especially if this rank ordering is stable over time. For
example, when we estimate a regression of log height on log ranking using the SHARE data, we get an R2 of
0.9813. This suggests that the rank ordering of countries based on height can almost predict the value of height
perfectly.
17
Hanushek and Woessman, 2008), economic literacy (e.g., Jappelli, 2010) and creativity (e.g.,
Florida and Tinagli, 2004). The results reported in this section are suggestive of the role that
height might play as a driver of policy related goals and the potential importance of ensuring
that growth is maximized during childhood.
6. Conclusions
Height has been used as a key marker of physical welfare and the standard of living,
as well as a marker of childhood health. Previous research based on national surveys has
found that height is correlated with cognitive function at older ages. Using data for 15
additional European countries, this paper provides further empirical support for the notion
that there exist a significant association between height and cognitive function in later life.
In this paper, we also ask a related and interesting question: are people from countries
where the average person is relatively tall smarter? To our knowledge, this paper is the first
to explore the link between height and cross-country differences in cognitive functioning.
Using data from the Survey of Health, Ageing, and Retirement in Europe, we find empirical
evidence that this is the case, even after controlling for self-reported childhood health, self-
reported childhood ability, parental characteristics and education.
In summary, the results of this paper suggest that average height is related to average
measures of cognitive functioning, both within a country and also when comparing across
countries. Furthermore, our results are also suggestive of the link between height and other
national indicators of interest, such as the cognitive functioning of high school students and
the creativity of the adult workforce. It therefore highlights the use of height as a potentially
useful national indicator. At present, height is not a statistic that is routinely collected during
censuses but there appears to be good reasons why there is potential value in doing so. A
shortcoming of our analysis is that our data do not allow us to systematically investigate why
there is a cross-country link between height and cognitive functioning. This would be an
interesting issue for further research.
18
References
Abbott, R., White, L., Ross, G., Petrovich, H., Masaki, K., Snowdon, D. and Curb, J. (1998).
Height as a marker of childhood development and late-life cognitive function: the Honululu-
Asia aging study. Pediatrics 102: 602–609.
Barber, N. (2005). Educational and Ecological Correlates of IQ: a Cross-National
Investigation. Intelligence, 33: 273–284.
Beard, A. and M. Blaser. (2002). The Ecology of Height: The Effect of Microbial
Transmission on Human Height. Perspectives Biology and Medicine, 45: 475–99.
Beauchamp, A., David Cesarini, Magnus Johannesson, Erik Lindqvist, Coren Apicella.
(2010). On the sources of the height–intelligence correlation: New insights from a bivariate
ACE model with assortative mating. Behavioral Genetics. DOI 10.1007/s10519-010-9376-7.
Berger, A. (2001). Insulin-Like Growth Factor and Cognitive Function. British Medical
Journal, 322 (January 27): 203.
Behrman, J, and M. Rosenzweig. (2004). Returns to Birthweight. Review of Economics and
Statistics. 86: 586–601.
Black, S., P. Devereux and K. Salvanes. (2007). From the Cradle to the Labor Market? The
Effect of Birth Weight on Adult Outcomes. Quarterly Journal of Economics, 122: 409–439.
Börsch-Supan, A. and Jürges, H. (eds). (2005). The Survey of Health, Ageing and Retirement
in Europe –Methodology. Mannheim: MEA.
Boström, G., and Diderichsen, F. (1997). Socioeconomic Differentials in Misclassification of
Height, Weight and Body Mass Index Based on Questionnaire Data. International Journal of
Epidemiology, 26: 860-866.
Bozzoli, C., A. Deaton and C. Quintana-Domeque (2009). Adult Height and Childhood
Disease. Demography, 46: 647–669.
Case, A. and C. Paxson. (2008a). Stature and Status: Height, Ability, and Labor Market
Outcomes. Journal of Political Economy, 116: 499–532.
Case, A. and Paxson, C. (2008b). Height, Health and Cognitive Function at Older Ages.
American Economic Review, 98: 463–467.
Case, A. and Paxson, C. (2009). Early Life Health and Cognitive Function in Old Age,
American Economic Review, 99: 104-109.
Cavelaars, A., A. E. Kunst, J. J. M. Geurts, R. Criaesi, L. Grotvedt, U. Helmert, E. Lahelma,
O. Lundberg, A. Mielck, N. KR. Rasmussen, E. Regidor, TH. Spuhler and J. P. Mackenbach.
(2000). Persistent variations in average height between countries and between socio-
economic groups: an overview of 10 European countries. Annals of Human Biology, 27: 407-
421.
19
Deaton, A. (2007). Height, Health, and Development. Proceedings of the National Academy
of Sciences, 104: 13232-13237.
De Groot, L., Sette, S., Zajkas, G., and Carbajal, A. (1991). Nutritional Status:
Anthropometry. Euronut SENECA Investigators. European Journal of Clinical Nutrition, 45:
31-42.
Duncan G., Brooks-Gunn J., Klebanov P. (1994). Economic Deprivation and Early
Childhood Development. Child Development, 65: 296–318.
Eppig, C., C. Fincher and R. Thornhill. (2010). Parasite Prevalence and the Worldwide
Distribution of Cognitive Ability. Proceedings of the Royal Society B: Biological Sciences
(doi:10.1098/rspb.2010.0973).
Eveleth, P. and Tanner, J. (1976). Worldwide Variation in Human Growth (Cambridge:
Cambridge University Press).
Eveleth, P. and Tanner, J. (1990). Worldwide Variation in Human Growth, 2nd edn
(Cambridge: Cambridge University Press).
Everson-Rose, S., Mendes de Leon, C., Bienias, J., Wilson, R., Evans, D. (2003). Early Life
Conditions and Cognitive Functioning in Later Life. American Journal of Epidemiology 158,
1083–1089.
Factor-Litvak, P. and Susser, E. (2004). A Life Course Approach to Neuropsychiatric
Outcomes, in (D. Kuh and Y. Ben-Shlomo, eds), A Life Course Approach to Chronic Disease
Epidemiology, 2nd edition, pp. 324–43, Oxford: Oxford University Press.
Florida, R. and I. Tinagli. (2004). Europe in the Creative Age. Demos Publishing.
Fogel, R. (1993). New Sources and New Techniques for the Study of Secular Trends in
Nutritional Status, Health, Mortality, and the Process of Aging. Historical Methods, 26: 5–43.
Garcia, J., Quintana-Domeque, C. (2007). The Evolution of Adult Height in Europe: a Brief
Note. Economics and Human Biology 5: 340–349.
Gelade, G. (2008). IQ, Cultural Values, and the Technological Achievement of Nations.
Intelligence, 36: 711-718.
Hanushek, E. and Woessmann L. (2008). The Role of Cognitive Skills in Economic
Development, Journal of Economic Literature, 46: 607–68.
Hatton, T and B. Bray. (2010). Long Run Trends in the Heights of European Men, 19th–20th
centuries. Economics and Human Biology, 8: 405–413.
Humphreys, L., T. Davey and R. Park. (1985). Longitudinal Correlation Analysis of Standing
Height and Intelligence. Child Development, 56: 1465–1478.
Jensen, A. (1983). Effects of Inbreeding on Mental-Ability Factors. Personality and
Individual Differences, 4: 71-87.
20
Jensen, A. and Sinha, S. (1993). Physical Correlates of Human Intelligence. In P. A. Vernon
(Ed.), Biological Approaches to the Study of Human Intelligence (pp. 139–242). Norwood,
NJ: Ablex.
Johnson, F. (1991). Biological Factors and Psychometric Intelligence: A Review. Genetic,
Social, and General Psychology Monographs, 117: 313–357.
Jappelli, T. (2010). Economic Literacy: An International Comparison. Economic Journal,
120: F429-F451.
Kanazawa, S. and Reyniers, D. (2009). The Role of Height in the Sex Difference in
Intelligence. American Journal of Psychology, 122: 527-536.
Kaplan G., Turrell G., Lynch J., et al. (2001). Childhood Socioeconomic Position and
Cognitive Function in Adulthood. International Journal of Epidemiology, 30: 256–63.
Kretchmer, N., J. Beard and S. Carlson. (1996). The Role of Nutrition in the Development of
Normal Cognition. American Journal of Clinical Nutrition, 63: 997S–1001S.
Lynn, R. (1989). A Nutrition Theory of the Secular Increases in Intelligence, Positive
Correlation between Height, Head Size and IQ. British Journal of Educational Psychology,
59 (November): 372–77.
Lynn, R. (1990). The Role of Nutrition in Secular Increases in Intelligence, 11: 273-285.
Lynn, R. (1991). The Evolution of Racial Differences in Intelligence. Mankind Quarterly, 32:
99–121.
Meenakshi F., G. Zamarro and E. Meijer. (2008). „Health Comparisons‟, in Health, Ageing
and Retirement in Europe – First Results from the Survey of Health, Ageing and Retirement
in Europe, eds. A. Börsch-Supan et al., 30-39. Mannheim: Mannheim Research Institute for
the Economics of Aging (MEA).
Palta, M., Prineas, R., Berman,R. and Hannan, P. (1982). Comparison of Self-Reported and
Measured Height and Weight. American Journal of Epidemiology, 115: 223-230.
Peck, A. and Lundberg, O. (1995). Short Stature as an Effect of Economic and Social
Conditions in Childhood. Social Science and Medicine, 41: 733–738.
Prince, M., Reischies, A., Beekman, R., Fuhrer, C., Jonker, S., Kivelä, B., Lawlor, A., Lobo,
H., Magnússon, I., Fichter, H., van Oyen, M., Roelands, I., Skoog, C. and J. Copeland.
(1999a). Development of the EURO-D scale – a European Union Initiative to Compare
Symptoms of Depression in 14 European Centres, British Journal of Psychiatry 174: 330-38.
Prince, M., Reischies, A., Beekman, R., Fuhrer, C., Jonker, S., Kivelä, B., Lawlor, A., Lobo,
H., Magnússon, I., Fichter, H., van Oyen, M., Roelands, I., Skoog, C. and J. Copeland.
(1999b). Depression Symptoms in Late Life Assessed using the EURO-D Scale. British
Journal of Psychiatry 174: 339-45.
21
Prokosch, M., Yeo, R. and Miller, G. (2005). Intelligence Tests with Higher G-loadings
Show Higher Correlations with Body Symmetry: Evidence for a General Fitness Factor
Mediated by Developmental Stability. Intelligence, 33, 203–213.
Rushton, J. (1995). Race, Evolution, and Behavior: a Life History Perspective. New
Brunswick, NJ: Transaction.
Saadat, M. (2008). Consanguinity and National IQ Scores. Journal of Epidemiology and
Community Health, 62: 566–567.
Silventoinen K., Posthuma D., van Beijsterveldt T., Bartels M, Boomsma D. (2006). Genetic
contributions to the association between height and intelligence: evidence from Dutch twin
data from childhood to middle age. Genes, Brain and Behavior, 5: 585–595.
Steckel, R. (1995). Stature and the Standard of Living. Journal of Economic Literature 33:
1903–1940.
Steckel, R. (2009). Heights and Human Welfare: Recent Developments and New Directions.
Explorations in Economic History, 46: 1–23.
Steward,A. (1982). The Reliability and Validity of Self-Reported Weight and Height. Journal
of Chronic Diseases, 35: 295-309.
Steward, A., Jackson, R., Ford, M. and Beaglehole, R. (1987). Underestimation of relative
weight by use of self-reported height and weight. American Journal of Epidemiology, 125:
122-126.
Sundet J. and Tambs K. (2005). Resolving the Genetic and Environmental Sources of the
Correlation between Height and Intelligence: A Study of Nearly 2600 Norwegian Male Twin
Pairs. Twin Research and Human Genetics, 8: 307–311.
Tanner, J. (1969). Relation of Body Size, Intelligence Test Scores and Social Circumstances.
In P. Mussen, J. Langer, & M. Covington (Eds.), Trends and Issues in Child Developmental
Psychology (pp. 182–201). New York: Holt, Rinehart, & Winston.
Templer, D. I. and Arikawa, H. (2006). Temperature, Skin Color, Per Capita Income, and IQ:
an International Perspective. Intelligence, 34: 121–139.
Van den Berg, G., D. Deeg, M. Lindeboom and F. Portrait. (2010). The Role of Early-Life
Conditions in the Cognitive. Decline due to Adverse Events Later in Life. Economic Journal,
120: F411-F428.
Wilson, R., P. Scherr, G. Hoganson, J. Bienias, D. Evans and D. Bennett. (2005). Early Life
Socio-Economic Status and Late Life Risk of Alzheimer's Disease, Neuroepidemiology, 25:
8–14.
Woodley, M. (2009). Inbreeding Depression and IQ in a Study of 72 Countries. Intelligence,
37: 268–276.
22
Zhang, Z., D. Gu, and M. Hayward. (2008) Early Life Influences on Cognitive Impairment
Among Oldest Old Chinese. Journal of Gerontology, 63B: S25–S33.
Table 1: Summary Statistics: Full sample
Variable Mean Std. dev. Min Max
Reading skill 3.66 1.15 1 5Writing skill 3.52 1.20 1 5Date questions score 3.81 0.54 0 4Verbal fluency 18.81 7.51 0 100Immediate recall score 4.89 1.86 0 10Delayed recall score 3.45 2.04 0 10Numeracy test score 0.62 0.24 0 1Depression ever 0.22 0.42 0 1Subjective health 0.26 0.44 0 1Euro-D score 7.82 2.12 0 10Height 167.73 9.12 100.68 220Age 65.01 10.49 50 105Male 0.44 0.49 0 1Foreign born 0.09 0.29 0 1Years of schooling 10.53 4.27 0 25Childhood health 2.06 1.01 1 5Father white collar when 14 0.26 0.44 0 1White collar job 0.52 0.49 0 1Childhood math 2.72 0.89 1 5Childhood language 2.69 0.87 1 5Childhood book 2.06 1.20 1 5
Notes: This table shows the summary statistics of variables for the respondents who were surveyed in the SHAREdata. Means are reported for the continuous variables and proportions are reported for categorical variables withthe standard deviations.
Table 2: Height, Health and Cognitive Functioning in SHARE
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Reading Writing Date Verbal Immediate Delayed Numeracy Depression Subjective Euro-Dskill skill questions fluency recall recall test ever health measure
score score score score
Height in cm 0.014 0.014 0.002 0.072 0.016 0.017 0.003 -0.001 0.003 0.013(17.34) (16.59) (4.82) (15.41) (14.14) (13.50) (17.85) (3.47) (8.71) (9.17)
Foreign born -0.195 -0.224 -0.034 -2.558 -0.368 -0.289 -0.033 0.021 -0.091 - 0.398(10.06) (11.10) (4.16) (22.84) (13.35) (9.62) (8.79) (3.14) (11.84) (11.53)
Age 0.025 0.027 0.063 0.246 0.109 0.064 0.006 -0.001 -0.008 0.137(4.68) (5.03 ) (18.12) ( 7.76) ( 14.33) (7.67) (5.48) (0.31) (3.96) (13.79)
Age2/100 -0.038 -0.042 -0.055 -0.339 -0.129 -0.099 -0.008 -0.002 -0.002 -0.122(9.49) (10.43) (20.27) (14.45) (22.70) (16.04) (9.36) (1.05) (1.15) (16.37)
Male -0.161 -0.188 -0.034 -0.328 -0.385 -0.501 0.035 -0.101 0.012 0.601(12.10) (13.27) (5.51) (4.19) (20.15) (23.50) (12.92) (22.63) (2.23) (25.33)
Wave dummies YES YES YES YES YES YES YES YES YES YESCountry dummies YES YES YES YES YES YES YES YES YES YESObservations 44649 44647 61939 61711 62029 62048 61255 62439 62879 62388R-squared 0.1774 0.1755 0.0719 0.2563 0.2269 0.2120 0.1246 0.0497 0.1018 0.1074
Notes: t-statistics are reported in parentheses and are in absolute values together with the coefficients which areestimated using OLS. Height is in cm. Standard errors are clustered at the individual level. SHARE waves 1-3are used covering the years 2002-2008 for 15 European countries.
23
Table 3: Coefficient on Height in SHARE: Table 1 for Sub-samples of 15 European Countries
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Reading Writing Date Verbal Immediate Delayed Numeracy Depression Subjective Euro-Dskill skill questions fluency recall recall test ever health measure
score score score score
Austria 0.011 0.014 0.003 0.094 0.026 0.020 0.002 -0.002 0.006 0.026(3.10) (3.94) (2.04) (3.46) (4.29) (2.92) (2.79) (1.33) (3.72) (3.96)
Germany 0.016 0.017 0.003 0.101 0.019 0.017 0.004 -0.001 0.003 0.009(6.29) (6.77) (2.64) (6.13) (5.08) (3.93) (8.06) (0.28) (2.31) (1.98)
Sweden 0.016 0.017 -0.001 0.091 0.011 0.008 0.003 -0.001 0.004 0.008(6.75) 6.97 (0.37) (5.22) (2.91) (1.75) (5.88) (0.35) (3.58) (1.76)
Netherlands 0.014 0.015 0.002 0.095 0.013 0.013 0.004 0.001 0.003 0.010(5.46) (5.53) (1.20) (6.76) (3.41) (2.93) (7.48) (0.36) (3.01) (2.22)
Spain 0.020 0.018 0.004 0.054 0.021 0.024 0.003 -0.002 0.004 0.022(5.07) (4.74) (2.36) (3.80) (4.96) (5.58) (4.82) (1.85) (3.25) (3.67)
Italy 0.026 0.025 0.003 0.129 0.031 0.029 0.004 -0.001 0.005 0.021(9.05) (8.40) (1.55) (7.97) (7.83) (6.45) (6.35) (0.11) (4.11) (3.75)
France 0.021 0.025 0.003 0.108 0.017 0.015 0.003 -0.001 0.002 0.012(7.10) (7.90) (2.01) (5.99) (4.11) (3.55) (5.53) (0.42) (1.41) (2.29)
Denmark 0.015 0.018 0.001 0.113 0.016 0.013 0.001 -0.002 0.004 0.012(5.14) (5.46) (0.15) (5.81) (3.46) (2.39) (5.21) (1.34) (2.65) (2.45)
Greece 0.015 0.014 0.001 0.039 0.019 0.019 0.002 -0.002 0.003 0.022(5.22) (3.51) (0.35) (2.94) (4.78) (4.64) (3.68) (3.08) (3.11) (4.87)
Switzerland 0.014 0.016 -0.001 0.062 0.012 0.012 0.002 -0.001 0.002 0.008(3.52) (3.52) (0.03) (2.25) (2.01) (2.09) (2.75) (0.97) (1.73) (1.46)
Belgium 0.012 0.012 0.001 0.079 0.017 0.022 0.003 -0.001 0.001 0.012(4.53) (4.63) (1.09) (5.75) (4.84) (5.34) (5.86) (1.53) (1.24) (2.65)
Israel 0.002 0.005 0.003 0.039 0.008 0.007 0.001 -0.002 -0.001 0.016(0.57) (1.39) (1.44) (2.01) (1.45) (1.47) (0.56) (1.78) (1.43) (2.36)
Czech Republic 0.010 0.011 0.006 0.065 0.021 0.021 0.003 -0.001 0.002 0.017(4.12) (4.25) (3.28) (3.42) (3.70) (4.20) (4.33) (0.17) (1.82) (2.58)
Poland 0.010 0.008 0.001 0.045 0.011 0.019 0.002 -0.002 0.001 0.005(2.68) (2.34) (0.68) (2.57) (2.26) (3.50) (3.09) (1.65) (0.42) (0.72)
Ireland 0.004 0.005 0.002 0.002 0.016 0.014 0.001 0.001 0.002 0.010(1.25) (1.57) (1.30) (0.11) (3.01) (2.26) (1.98) (0.59) (1.52) (2.06)
Notes: t-statistics are reported in parentheses and in absolute values together with the coefficients which areestimated using OLS. Height is in cm. Standard errors are clustered at the individual level. SHARE waves 1-3are used covering the years 2002-2008 for 15 European countries.
24
Table 4: Determinants of Height, Education, and Occupational Choice in SHARE
(1) (2) (3) (4)Height Years of White collar White collarin cm schooling job job
Height in cm 0.054 0.005 0.002(11.34) (7.34) (3.47)
Years of schooling 0.044(41.40)
Father white collar 1.216 2.284 0.305 0.206job when 10 (12.08) (35.62) (35.75) (23.85)Childhood health 0.133 0.253 0.007 0.006
(1.38) (4.43) (0.81) (0.79)Foreign born -1.268 -0.135 -0.037 -0.037
(6.70) (1.10) (2.26) (2.41)Age -0.107 -0.175 0.020 0.029
(2.07) (5.79) (3.88) (5.87)Age2/100 -0.022 0.053 -0.017 -0.020
(0.56) (2.35) (4.88) (5.87)Male 11.402 0.360 -0.140 -0.158
(128.77) (4.80) (13.54) (16.44)Wave dummies YES YES YES YESCountry dummies YES YES YES YES
Notes:t-statistics are reported in parentheses and in absolute values together with the coefficients which areestimated using OLS. Height is in cm. Standard errors are clustered at the individual level. SHARE waves 1-3are used covering the years 2002-2008 for 13 European countries.
Table 5: Childhood and Mid-life Circumstances in SHARE
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Reading Writing Date Verbal Immediate Delayed Numeracy Depression Subjective Euro-Dskill skill questions fluency recall recall test ever health measure
score score score score
Height in cm 0.010 0.011 0.002 0.054 0.012 0.013 0.002 -0.001 0.002 0.012(14.08) (13.72) (4.05) (12.03) (10.62) (10.59) (14.64) (2.85) (6.40) (7.59)
White collar 0.376 0.444 0.025 1.365 0.386 0.389 0.054 -0.004 0.076 0.223job (22.95) (25.99) (3.74) (14.86) (16.97) (14.58) (17.41) (0.62) (10.60) (6.27)Father white collar 0.256 0.280 0.005 1.079 0.249 0.249 0.030 0.002 0.040 0.060job when 14 (17.83) (18.52) (1.03) (12.12) (11.91) (10.08) (10.75) (0.29) (6.61) (1.94)Childhood health 0.143 0.156 0.016 0.302 0.070 0.016 0.013 -0.045 0.096 0.340
(10.35) (10.94) (2.87) (3.76) (3.58) (0.72) (4.68) (8.80) (16.12) (13.02)Years of schooling 0.070 0.074 0.007 0.293 0.066 0.068 0.009 -0.001 0.010 0.035
(36.64) (37.18) (10.39) (32.42) (30.48) (28.25) (30.50) (2.31) (18.03) (11.45)Controls YES YES YES YES YES YES YES YES YES YESWave dummies YES YES YES YES YES YES YES YES YES YESCountry dummies YES YES YES YES YES YES YES YES YES YES
Notes: t-statistics are reported in parentheses and in absolute values together with the coefficients which areestimated using OLS. Height is in cm. Standard errors are clustered at the individual level. SHARE waves 1-3are used covering the years 2002-2008 for 13 European countries.
25
Table 6: Determinants of Height, Education, and Occupational Choice in SHARE with Addi-tional Controls for Childhood History and Parent Characteristics
(1) (2) (3) (4)Height Years of White collar White collarin cm schooling job job
Height in cm 0.036 0.062 0.046(18.23) (15.75) (10.54)
Childhood conditionsMath ability 0.331 0.905 0.083 0.053
(6.63) (33.57) (19.28) (12.49)Language ability 0.299 1.048 0.104 0.066
(5.74) (37.37) (23.25) (14.88)Books 0.536 0.999 0.087 0.047
(12.68) (42.57) (23.58) (12.38)Parents drank heavily -0.549 -0.869 -0.090 -0.043
(3.51) (10.75) (6.76) (3.41)Parents had mental health problems -0.737 0.099 0.021 0.011
(2.51) (0.65) (2.52) (1.47)Father present at home 0.287 0.245 0.025 0.016
(1.82) (2.76) (1.83) (1.24)Mother present at home 0.553 0.321 0.039 0.015
(2.43) (2.47) (1.91) (0.78)Dwelling with no features -0.756 -1.366 -0.162 -0.112
(6.59) (22.62) (16.89) (12.11)Dwelling with > 3 rooms 0.399 0.756 0.055 0.025
(3.96) (13.28) (6.21) (2.99)Dwelling with more than 4 people -0.546 -0.569 -0.061 -0.038
(5.94) (11.14) (7.35) (4.93)Parent’s characteristics
Mother’s age at death 0.009 0.013 0.003 0.001(2.77) (4.72) (4.42) (1.41)
Father’s age at death 0.007 0.008 0.001 0.001(2.11) (2.86) (1.02) (0.49)
Mother’s current health 0.200 0.155 0.012 0.007(2.84) (3.26) (1.62) (0.81)
Controls in Table 4 YES YES YES YESWave dummies YES YES YES YESCountry dummies YES YES YES YES
Notes: t-statistics are reported in parentheses and in absolute values together with the coefficients which areestimated using OLS. Height is in cm. Standard errors are clustered at the individual level. SHARE waves 1-3are used covering the years 2002-2008 for 13 European countries.
26
Table 7: The Role of Childhood and Mid-life Circumstances in SHARE with Additional Controlsfor Childhood History and Parent Characteristics
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Reading Writing Date Verbal Immediate Delayed Numeracy Depression Subjective Euro-Dskill skill questions fluency recall recall test ever health measure
score score score score
Height 0.009 0.010 0.001 0.046 0.010 0.014 0.002 -0.001 0.005 0.009in cm (12.94) (12.47) (3.84) (10.54) (9.58) (7.40) (13.56) (2.81) (7.64) (6.91)ChildhoodconditionsMath 0.150 0.172 0.017 0.870 0.182 0.207 0.041 -0.005 0.022 0.167ability (20.74) (23.04) (5.96) (20.81) (18.10) (17.32) (29.51) (1.75) (7.33) (10.70)Language 0.253 0.287 0.012 0.868 0.217 0.221 0.025 0.009 0.020 0.073ability (34.50) (37.89) (4.19) (19.64) (20.55) (17.71) (17.15) (3.45) (6.37) (4.54)Books 0.114 0.125 0.003 0.841 0.137 0.141 0.017 0.004 0.017 0.050
(19.41) (20.24) (1.45) (23.54) (15.97) (14.18) (14.96) (1.89) (6.64) (3.92)
Parents -0.001 -0.016 -0.013 -0.108 -0.011 -0.011 -0.005 0.019 -0.027 -0.036smoked (0.05) (1.19) (2.50) (1.44) (0.58) (0.49) (2.04) (4.18) (2.68) (3.39)Parents drank -0.091 -0.112 -0.041 -0.419 -0.153 -0.130 -0.024 0.054 -0.113 -0.468heavily (3.85) (4.60) (4.05) (3.15) (4.69) (3.46) (5.60) (6.22) (6.00) (8.84)Parents mental -0.059 -0.080 -0.037 -0.497 -0.024 -0.034 -0.026 0.149 -0.105 -0.589health problems (1.44) (1.84) (2.49) (1.92) (0.39) (0.47) (3.06) (8.27) (3.04) (6.12)
Mother present 0.287 0.051 0.007 -0.058 0.029 0.071 0.011 -0.032 0.052 0.178at home (0.47) (1.47) (0.49) (0.29) (0.69) (1.34) (1.76) (2.67) (1.94) (2.36)Father present 0.001 0.018 0.017 -0.012 0.019 0.043 0.014 -0.029 0.051 0.154at home (0.3) (0.77) (1.74) (0.09) (0.58) (1.18) (3.22) (3.48) (2.68) (3.06)
Dwelling with no -0.161 -0.170 0.002 -0.630 -0.188 -0.152 -0.015 0.011 -0.031 0.021features (10.04) (10.26) (0.33) (7.47) (8.69) (6.05) (5.03) (1.96) (2.49) (0.59)Dwelling with 0.070 0.075 0.009 0.692 0.113 0.057 0.015 -0.004 0.040 0.114> 3 rooms (5.08) (5.28) (1.58) (8.74) (5.91) (2.57) (5.77) (0.88) (3.68) (3.87)Dwelling with -0.080 -0.082 -0.001 -0.159 -0.079 -0.084 -0.009 0.008 -0.010 -0.060> 4 people (6.21) (6.03) (0.18) (2.07) (4.29) (3.91) (3.56) (1.61) (1.06) (2.15)
Mother’s age 0.002 0.002 0.001 0.011 0.003 0.003 0.001 0.001 0.002 0.004at death (4.04) (4.48) (3.92) (4.28) (5.11) (4.74) (2.21) (0.15) (6.04) (4.33)Father’s age 0.001 0.001 0.001 0.005 0.002 0.001 0.001 -0.001 0.002 0.003at death (0.48) (0.99) (0.54) (2.30) (2.43) (0.20) (3.31) (4.52) (6.02) (3.49)
Mother’s current 0.075 0.081 0.001 0.116 0.051 0.063 0.010 -0.016 0.003 0.197health (8.5) (8.64) (0.53) (2.16) (3.98) (4.25) (5.90) (4.53) (8.54) (10.29)Father’s current 0.049 0.074 0.001 0.143 0.045 0.039 0.001 -0.026 0.003 0.157health (3.81) (5.49) (0.26) (1.80) (2.37) (1.76) (0.31) (4.94) (8.54) (5.28)
Controls in Table 5 YES YES YES YES YES YES YES YES YES YESWave dummies YES YES YES YES YES YES YES YES YES YESCountry dummies YES YES YES YES YES YES YES YES YES YES
Notes: t-statistics are reported in parentheses and in absolute values together with the coefficients which areestimated using OLS. Height is in cm. Standard errors are clustered at the individual level. SHARE waves 1-3are used covering the years 2002-2008 for 13 European countries.
27
Table 8: Coefficients on Country Dummies in the Height, Cognitive Functioning and HealthRegressions
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)Height Reading Writing Date Verbal Immediate Delayed Numeracy Depression Subjective Euro-D
skill skill questions fluency recall recall test ever health measurescore score score score
Austria 5.8899 1.0787 1.0741 0.2014 7.1517 1.5905 1.1720 0.1654 0.1358 0.3356 0.9353Germany 6.5708 0.7466 0.6990 0.1749 5.6868 1.5976 1.0033 0.1630 -0.0958 0.1024 0.8661Sweden 7.4292 1.4068 1.3788 0.2018 7.9109 1.5187 1.3571 0.1621 -0.0517 0.4269 0.8531Netherlands 8.1212 0.6006 0.5157 0.1344 4.4244 1.3572 1.0862 0.1571 -0.0444 0.2980 0.7974Italy 2.0260 0.3444 0.3028 0.1306 -0.2433 0.5336 0.2646 0.0432 -0.0449 0.0131 0.0469France 3.0406 0.8693 0.7547 0.1431 4.5292 0.7446 0.4641 0.0904 -0.0076 0.1333 0.1263Denmark 7.3879 0.9211 0.8747 0.1183 5.8501 1.4633 1.3197 0.1262 -0.0343 0.4191 0.8721Greece 3.5556 0.5080 0.4713 0.2169 -0.5086 0.9941 0.6415 0.1321 -0.1387 0.3077 0.8219Switzerland 5.0631 1.0420 0.9573 0.1967 5.3617 1.4749 1.1688 0.1762 -0.0831 0.5839 0.9543Belgium 4.4445 0.9062 0.7786 0.1188 4.3370 1.0458 0.6041 0.0861 -0.0295 0.2582 0.4362Israel 4.0857 0.7993 0.8373 0.0581 4.7328 1.0076 0.5612 0.1463 -0.1125 0.2056 0.2430Czech 5.6134 0.7706 0.7307 0.0842 2.7359 0.8619 0.2280 0.1079 0.0951 -0.0685 0.5936Poland 2.8888 0.4665 0.4602 0.1172 0.3462 0.4060 0.0669 0.0496 -0.0336 -0.4911 -0.8795Ireland 1.5199 1.0363 0.8621 0.1408 0.4046 1.2848 1.2528 0.0866 -0.0137 0.5592 0.8139
Notes: Column 1 are the estimated coefficients on country dummies in Table 6 column 1. Columns 2-10 are theestimated coefficients on country dummies in Table 7. For Ireland and Israel, regressions are estimated withoutthe additional controls given in Table 6 that are taken from SHARELIFE.
Table 9: Cross-Country Correlations between Height and Cognitive Functioning and betweenHeight and Health
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Reading Writing Date Verbal Immediate Delayed Numeracy Depression Subjective Euro-Dskill skill questions fluency recall recall test ever health measure
score score score scoreHeightPearson 0.3670 0.4225 0.1679 0.7538 0.6818 0.5442 0.7231 0.0929 0.2576 0.5396Spearman’s rank 0.3363 0.3890 0.1736 0.7143 0.7011 0.5121 0.6791 -0.0945 0.2571 0.5824Kendall’s rank 0.2747 0.2967 0.1209 0.5385 0.5165 0.4286 0.4725 -0.0110 0.2308 0.4066
Notes: Each row presents the correlation coefficients between column 1 and one of the columns 2-10 in Table 8.
Table 10: Possible National Benefits of National Height
Coefficients of correlation between the country dummies in height and national indicators.
National Indicator Pearson Spearman’s rank Kendall’s rank
Average mean literacy score (PISA 2009) 0.4975 0.4209 0.2949Average mean mathematics score (PISA 2009) 0.4583 0.5165 0.3407Creativity Index 0.8171 0.8545 0.6889Tolerance Index 0.8555 0.903 0.7778Technology Index 0.6911 0.7818 0.5556Talent Index 0.5142 0.503 0.3778
Notes: Each row presents the correlation coefficients between the corresponding national variables and column 1in Table 8. PISA scores are obtained from the OECD database, and the creativity indices are from Florida andTinagli (2004).
28
(a) Full sample (b) Men (c) Women
Figure 1: Average Height by Country: 15 European Nations
(a) Full sample (b) Men (c) Women
Figure 2: Average Numeracy Score by Country: 15 European Nations
(a) Full sample (b) Men (c) Women
Figure 3: Average Dates Score by Country: 15 European Nations
29
(a) Full sample (b) Men (c) Women
Figure 4: Average Verbal Fluency Score by Country: 15 European Nations
(a) Full sample (b) Men (c) Women
Figure 5: Average Immediate Recall Score by Country: 15 European Nations
(a) Full sample (b) Men (c) Women
Figure 6: Average Delayed Recall Score by Country: 15 European Nations
30
(a) Height and Immediate Recall (b) Height and Delayed Recall
(c) Height and Numeracy (d) Height and Verbal Fluency
(e) Height and Date Questions (f) Height and Euro-D Measure of Wellbeing
Figure 7: Positive Correlation between Height and Cognitive Functioning: 15 European Nations
31