The effect of the seasons of the year on malnutrition in North Korea

ARTICLE IN PRESS l l HOMO — Journal of Comparative Human Biology 60 (2009) 59–75

www.elsevier.de/jchb

The effect of the seasons of the year on malnutrition in North Korea D. Schwekendiek Chair of Economic History, Department of Economics, University of Tuebingen, Mohlstr. 36, D-72074 Tuebingen, Germany Received 29 November 2006; accepted 10 March 2008

Abstract North Korea’s economic isolation as a consequence of its recent nuclear testing could lead to another famine. In this context, the article investigates health determinants of the last famine. Birth season is a reliable proxy for nutritional and epidemiological circumstances in early life, which in turn can systematically influence later-life health outcomes such as stature. The aim of this study is to assess the effect of birth seasons on the development of heights. Height-for-age z-scores for up to 9934 pre-school children measured in 1997 during the North Korean food crisis in the decade of the 1990s and measured in 2002, are used for the analysis. Using these data, we have verified a number of earlier studies showing that cohorts born in winter are smaller. We have also found that cohorts born in autumn are taller. Additionally, when we control for sex, birthplace, and birth year, these calendar effects remain robust. Ignoring the humanitarian dilemma of any sort, it may be concluded that in totalitarian North Korea, Pyongyang can easily influence the biological welfare of its people by interfering with the birth calendar, in order to reduce the magnitude of a potentially upcoming famine. r 2008 Elsevier GmbH. All rights reserved.

Tel.: +49 7071 29 78160; fax: +49 7071 29 5119.

E-mail address: [email protected]. 0018-442X/$ - see front matter r 2008 Elsevier GmbH. All rights reserved. doi:10.1016/j.jchb.2008.03.001

ARTICLE IN PRESS 60

D. Schwekendiek / HOMO — Journal of Comparative Human Biology 60 (2009) 59–75

Introduction Totalitarian North Korea recently became a nuclear super-power. Even though it is now almost daily in the media, there is a lack of information on the common people living in the country. In this paper, we use height measurements of North Koreans, which offer reliable information on general living conditions of the North Korean people. Stature is an established and safe indicator of the standard of living particularly in developing and less-developed countries, and in historical societies (Fogel, 1983; Komlos and Baten, 1998). It is likewise an established procedure used by the United Nations to assess the overall health status within a population using anthropometric measures (De Onis and Blo¨ssner, 1997). Furthermore, in a totalitarian context, stature has an advantage over demographic or economic indicators in which it cannot be directly manipulated, and neither is a comprehensive theoretical framework required for obtaining meaningful results. Human growth is largely determined by socio-economic conditions in early life, so that height measurements reflect the nutritional, epidemiological, and social stress populations have been exposed to (Tanner, 1990). (Here, it is important to remember that for this kind of assessment heights are taken into account on an average, since on an individual level, human growth is largely controlled by genes) In this context height measurements reflect an average in the population. What has determined the stature of North Koreans in the past? In the 1990s, North Korea’s trade system thoroughly disintegrated. This was the consequence of geopolitical transformations following the collapse of the Eastern Bloc (when Pyongyang suddenly had to pay hard currency for formerly subsidised goods) and, of Pyongyang inefficient central-planning economy. Prior to the crisis, substantial imports consisted of food, fertilizers, and energy. With the economic breakdown, living standards deteriorated as well. The government was no longer able to feed its 20 million people, nor was it able to earn foreign money to purchase food from the world market, given that the energy crisis stalled industrial production. Furthermore, in 1995 and 1996, the country was seriously struck by natural disasters that triggered one of the most terrible famines in modern history. Since then, North Korea has been receiving international food aid. According to the latest survey, which was conducted in 2004, there is still a high prevalence of chronic malnutrition among North Korean children (CBS, 2004). Even in 2005, North Korea was the second largest recipient in the world of United Nations food donations, suggesting that the food crisis is not yet over. In addition, the current economic sanctions imposed by the international community in response to North Korea’s recent nuclear testing, could aggravate the situation for the people even further. For all these reasons, research on the biological welfare of North Koreans has certainly a political dimension as well. In this context, Schwekendiek (2008) analysed height measurements of North Korean children collected in 1997. Stature was linked to a large set of socioeconomic factors employing demographic data officially released by the government, as well as cartographic data. In particular, a number of hypotheses that developed with time and had accumulated in the literature, were quantitatively tested and

ARTICLE IN PRESS D. Schwekendiek / HOMO — Journal of Comparative Human Biology 60 (2009) 59–75

61

discussed in socio-economical terms. These included the ‘military first’, the ‘triage’, the ‘farmers’ markets’, and the ‘flood damage’ hypotheses. The role of the public distribution system (PDS) and the question of whether rural or urban areas are better-off were also investigated. As one of the results, local harvest conditions were found to have a significant and consistent effect on the living standard of North Koreans. It has been concluded that the quantity and quality of food supplied through farmers’ grey- and black markets (existing semi-legally in every North Korean administrative county) determine human well-being in this country. A drawback in this analysis was that most of the exogenous variables had to be calculated on a county-administrative level, as no disaggregated information was available for these sets of variables. The aim of this paper is to investigate the effect of birth seasons on the North Korean standard of living during the food crisis of the decade in the 1990s. In general, birth period is an established proxy for seasonally varying early-life factors that systematically determine later-life health outcomes (this will be discussed further in the next section). This paper deliberately neglects the socio-economic effects analysed by Schwekendiek in 2008 and instead opts for a biological approach where basic variables, directly included in the dataset, are mainly used. These basic variables comprise sex, location, and birth date—in terms of the birth year and season of birth. As recommended by the United Nations, these are standard variables to be collected during all surveys along with the height data.

Previous studies Why does the birth season have an effect on our biological characteristics? There can be no doubt that certain sections of the solar calendar are associated with seasonally varying phenomena like the duration of sunshine, the amount of precipitation, or average temperature. These, again, have a large impact on the seasonal variation of food crops and of air- and waterborne diseases. Thus, it can be assumed that nutritional and epidemiological conditions vary largely throughout the year. Yet, as shown by Henneberg and Louw (1993), seasonal effects seem to be still present in dogs whose nutrition did not vary throughout the year. Thus, it may be concluded that seasonally varying quantity and quality of food are not the only factors that seem to cause seasonal effects on health. However, one has to consider the nutritional interaction with diseases, as outbreaks of many infections can be due to nutritional deficits (Baten and Wagner, 2003). In this light, seasonally varying nutritional factors, disease environment, and most likely a combination of both are likely to have caused these seasonal patterns. In the debate on ‘superstition versus science’, several hundred scientific studies on a variety of human welfare indicators have revealed the existence of birth month patterns. A discussion of each proposed indicator, let alone study, would be far beyond the scope of this paper. For a systematic overview, see Doblhammer (2004). A comprehensive survey of the literature can also be found in Axt and AxtGadermann (2004). The present paper only introduces some generally important

ARTICLE IN PRESS 62

D. Schwekendiek / HOMO — Journal of Comparative Human Biology 60 (2009) 59–75

studies and discusses those that specifically investigate the stature in the context of birth periods. Research in this area was pioneered by Huntington (1938). In the book ‘‘Seasons of Birth’’ he established a link between the calendar seasons and socio-psychological outcomes. From then on, the indicator birth month has been extensively employed and established in psychology where mostly seasonal effects on the risk of mental diseases were studied. Thus, the majority of studies that established the link between birth seasonality and health outcomes came from epidemiologists. In the review of over 250 studies, winter–spring birth cohorts have been found more frequently (5–8%) to suffer from schizophrenia or mania (Torrey et al., 1997). However, recently, a considerable number of studies have been published also by demographers (Doblhammer and Vaupel, 2001) and anthropometrists (Banegas et al., 2001; Chenoweth and Canning, 1941; Mills, 1941; Philips and Young, 2000; Weber et al., 1998; Wohlfahrt et al., 1998; Xu et al., 2001). Current studies are anchored in the groundbreaking works of Doblhammer, 2004. In her publication entitled ‘‘The Late Life Legacy of Very Early Life’’ she discusses life expectancy in the context of birth months. With regard to Denmark, Austria, USA, and Australia, Doblhammer provides evidence that the birth month is statistically linked to differences in lifespan. Moreover, she shows that the calendar seasons of the year do matter by comparing seasonal birth patterns in lifespan in the northern hemisphere to those found in the southern hemisphere—given that the harvest season is meteorologically reversed there. Moreover, looking at UK migrants to Australia, Doblhammer observes that their birth month pattern in life expectancy resembles that of their home country in the northern hemisphere, thus providing evidence that the birth month seems to be systematically determine health. As a main result, Doblhammer reports that autumn cohorts—i.e. babies born during harvest season—have statistically the highest life expectancy. On the other hand, this effect is rather weak, as a difference in lifespan of only 6 months, at the most, was found. One frequently raised concern is that beyond early-life factors, life-course factors have an impact on biological development as well. The birth month concept does allow for catch-up growth and also purports that any negative shock in later life can lower an individual’s health status. Yet on an average, when considering a larger number of observations, systematic differences by birth month do remain. Seasonal birth patterns in health are even prevalent in modern societies (Weber et al., 1998; Wohlfahrt et al., 1998) where the nutritional and medical environment usually plays a minor role. Summarising, it is now widely accepted that the birth period is an indicator of seasonally varying early-life factors, which, again, are supposed to have a statistical impact on further biological development of individuals. Previous studies concerning the Northern hemisphere, that linked birth period to stature, are reported in Table 1—data are based on Doblhammer (2004), pp. 162–166. Note that some studies focus on seasonal findings, whereas the majority of them focus on monthly findings. In this table as well as in the following, as a matter of causality, we refer to meteorological seasons that slightly differ from the traditional calendar classifications. What is particularly striking is that the cohorts born in spring in the Northern hemisphere seem to be the tallest, whereas the cohorts

ARTICLE IN PRESS

Notes: Dark colour represents maximum heights, light colour represents minimum heights.

D. Schwekendiek / HOMO — Journal of Comparative Human Biology 60 (2009) 59–75

Table 1. Studies on stature and birth season in the northern hemisphere.

63

ARTICLE IN PRESS 64

D. Schwekendiek / HOMO — Journal of Comparative Human Biology 60 (2009) 59–75

born in winter are the shortest. This correlation seems to be robust—not only across different societies, but also across different age groups that range in stature from length-at-birth to terminal stature. Beyond this, the results for the other seasons are rather mixed, which is discussed below in the North Korean context. Henneberg and Louw (1990) conducted studies on the heights of children in South Africa, thus being raised in the Southern hemisphere. They found that South African cohorts born in September–November are the tallest (corresponding to the meteorological spring in the Southern hemisphere), and those born in February–July (corresponding to the meteorological autumn and winter in the Southern hemisphere) to be the smallest. Thus, in both hemispheres, cohorts born in spring seem to be the tallest, whereas those born in late autumn and winter seem to be shorter. In other terms, the birth period effects on general health measured by height, seem to be mirrored and meteorologically shifted in the two hemispheres. Henneberg and Louw (1990) concluded that ‘‘the position of our planet with respect to the sun’’ (p. 232) might have caused this result. As the eccentricity of the earth’s orbit is not supposed to affect the seasons, their result clearly casts doubt that nutritional variations in early life, specifically the timing of the harvest, have an impact. In this study, birth season is regarded as a proxy for early life factors. Thus, we do not seek to investigate the underlying dynamics, or causes of seasonal variations in heights. As mentioned before, human growth is a very complex process which is subject to epidemiological and nutritional influences beyond genetic predisposition. It is also largely subject to the sensitive periods of this process—concerning both preand post-natal stages. It can be speculated that winter cohorts are more likely to become stunted because some diseases, notably respiratory infections, are more frequent in winter. This could partially explain why these studies find that the tallest cohorts are born in spring, as these cohorts are exposed to a better overall environment during spring and summer in their first months of life. On the other hand, the fact that these cohorts suffered more during their last three months in utero would be ignored; that is, for instance, the critical time when the human immune system develops. It is important to note that the indicator birth season does not allow the investigation of these dynamic mechanisms—nor is it geared towards this—it is simply a statistical reference that enables us to see systematic differences by birth. Besides the nutritional and epidemiological factors other biological factors might be of interest that are associated with birth season. For instance, Centola and Eberly (1999) have shown that the quality of sperm varies throughout the year, and Weber et al. (1998) suggested that sunshine durations trigger hormonal production. Thus, nutritional, epidemiological, and biological early-life factors, and perhaps a combination of these, can be responsible for these anthropometric effects.

Materials and methods This study uses survey data collected in the years 1997 and 2002 by the United Nations in collaboration with North Korean authorities. These surveys have been

ARTICLE IN PRESS D. Schwekendiek / HOMO — Journal of Comparative Human Biology 60 (2009) 59–75

65

Ru

China

ssi

a

reviewed extensively elsewhere (Katona-Apte and Mokdad, 1998; Schwekendiek, 2007). Thus, in this study only some basic characteristics are presented. The first systematic survey of that kind in North Korea was carried out in 1997 by Katona-Apte and Mokdad (1998). This survey coincided with the peak of the North Korean food crisis. The United Nations and its agencies were allowed to measure the heights of 3942 pre-school children in 40 institutions and 19 counties representing 5 provinces. In 2002, the United Nations collected data on 6000 pre-school children representing 63 counties in 10 provinces (CBS et al., 2002; Schwekendiek, 2007). Height data were collected at the household level, not at the institution level as in the 1997 institution-based survey. For an overview of the North Korean administrative provinces, see Fig. 1. With the exception of the northern province Chagang (a mountainous region over 2000 m above sea level, where implementation of surveys presumably became too complicated) 11 out of 12 North Korean administrative provinces are represented in these surveys and then in the analysis.

North Korea

g on gy m Ha th r No

n

pa

Ja

South Korea an

Oce Pacific

Ryanggang Chagang

South Hamgyong North Pyongan

South Pyongan ang

ngy

Pyo

Nampo

Kangwon

North Hwangae South Hwangae

g on es a K

Fig. 1. North Korean provinces.

ARTICLE IN PRESS 66

D. Schwekendiek / HOMO — Journal of Comparative Human Biology 60 (2009) 59–75

In this study, the 1997 and 2002 datasets are combined and then a pooled crosssectional regression is made. In order to show that anthropometric seasonal effects remain consistent for both surveys the datasets for each survey are also examined separately. The 1997 and 2002 children’s height measurements are related to the WHO/NCHS reference curves (Waterlow et al., 1977) by using height-for-age z-scores (HAZ). In this formula the h denotes a height measurement and k denotes a measurement group, defined by age and sex. Thus, the HAZ-value of an individual i is expressed as HAZ i ¼

hik  hk , sk

where h¯k is the median height and sk is the standard deviation of height in the corresponding group that is defined by the age and sex of the WHO/NCHS reference population. If the absolute height measurements hi would have been used, then, the sub-sample size will be sometimes limited to only a single observation, because heights of infants and pre-school children must be systematically differentiated by sex and age in each month. Hence, by employing relative and standardised HAZ, the data for the entire sample can be used in a straightforward analysis. In order to see the additional biological and social impact of these variables during the North Korean famine, control dummies for sex and birth cohorts are introduced in the regression analysis. From the statistical point of view, HAZ scores are superior to absolute height measurements for the analysis of stature, and, in addition, this method is strongly recommended by the United Nations. Linear regressions are used to show that these effects are statistically significant, and also to conduct a multivariate analysis. In addition, a set of dummies for the birth seasons is introduced to capture the seasonal effects and to see the sign and significance level of the respective coefficients. This technique is widely applied in research on birth periods and health (Doblhammer, 2004; Ferry et al., 2006).

Results and discussion Table 2, shows an average HAZ scores by birth season. These scores are based on measurements of 9934 pre-school children born in the years 1991–2002. Fifty observations were removed because of data entry errors. Contrary to previous studies on height and birth season (Table 1), the preliminary result is that, on an average, children born in autumn seem to be the tallest. This is indicated by HAZ of 1.61 (Table 2). Conversely, cohorts born in winter and spring are the shortest, as shown by the highest negative HAZ score. Also, note that irrespective of the seasons, all HAZ values have a negative sign and are approximately close to 1.7. This result clearly indicates that all sub-cohorts have experienced continuous and considerable nutritional stress. The negative winter effect seems to be consistent with previous studies on stature and birth period (Table 1). However, many of these studies likewise report a positive spring effect,

ARTICLE IN PRESS D. Schwekendiek / HOMO — Journal of Comparative Human Biology 60 (2009) 59–75

67

Table 2. Health and food seasonality in North Korea. Seasons Crop calendar

Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb

Birth calendar

Spring

Summer

Autumn

Winter

1.75

1.67 H H H

1.61

1.75

Height for Age (HAZ) Wheat Spring potato Rice Maize Sweet potato Soya bean Main season potato

H

H H H H

H H

Source: crop calendar based on Palka and Galgano (2004). H ¼ Harvest.

which is not seen in this study. Quite the opposite, spring cohorts seem to be as disadvantaged as winter cohorts. Furthermore, only some of the previous studies reported a positive (early) autumn effect (Cole, 1993; Otto and Glaass, 1959; Otto and Noack, 1957). Seasonal improvement in heights seems to be correlated to the North Korean harvest calendar (Table 2). In general, the harvest is brought in around summer and early autumn. From May to July, the first crops, that is only wheat and spring potatoes, are harvested. The autumn particularly matters, as most crops are harvested at that time including the main staple crops in North Korea, that is rice and maize. The main vegetable, cabbage, is also harvested in late autumn. Considering the North Korean national food deficit, it may be possible that the winter cohorts have not been able to benefit from the seasonal peaks regarding both the quantity and quality of food. This may also explain why the situation is as bad in spring as it is in winter and why it slightly improves in the summer when the first harvest is brought in. This suggests that food seasonality could have had an impact on the development of stature. However, harvested foodstuffs might not directly have had an effect in early life as newborns are largely breastfed, where the quality and quantity of the mother’s milk primarily depend on her energy stores built over long periods of time. Therefore, the nutritional status of newborns should only be slightly, if at all, influenced by the immediate nutritional intake of the mothers. This requires further investigations, specifically considering that disease environment or sunshine durations etc. can be, likewise, expected to correlate with the harvest calendar. In Table 3, regression 1 represents the basic model, where only the seasonal effect dummies are included. As seen, the spring and winter coefficients are statistically significant, and the sign and the coefficients mirror the descriptive results shown in Table 2, that cohorts born in winter and spring are significantly shorter than those in

ARTICLE IN PRESS 68

D. Schwekendiek / HOMO — Journal of Comparative Human Biology 60 (2009) 59–75

the summer. Also note that the autumn dummy is close to the critical significant level, thus, cohorts born in autumn are taller than those born in the summer at the 8% level. This coefficient clearly turns significant at the 1% level once we control for further relevant variables (Table 3, regressions 2–4). The overall impact of the birth season seems to be rather small, as the basic model can explain only 0.02% of the variations in HAZ, but as indicated by the joint F-test statistic, birth season per se can statistically indicate the growth status of the children. In regression 2 further variables are incorporated from the datasets. This is to see whether variations by birth year, sex, and location have a systematic influence on the development of heights. It has been found that males seem to have suffered more during the North Korean food crisis. This result is consistent with previous research on famines (Macintyre, 2002) and seems to be due to the genetic constellation of females who are biologically more robust to dire conditions. Also, as noted by Goodkind (1999), there is no evidence of sex preference in communist North Korea. Hence, it may be concluded that this result is largely due to human biological factors and has not been confounded by social factors. In addition, birth year dummies are introduced to see whether annual time trends matter for the seasonal effects. Based on regressions 2 and 3 (Table 2), it seems that cohorts born before 1995 (i.e. cohorts born in 1991 and 1993) are smaller compared to those born in the reference year or later. This finding is corroborated by regression 4 (Table 3). Cohorts born after 1995 seem to be taller than those of the reference year (though only statistically significantly taller for the year 2002). It may be reasonably assumed that the year 1995 represents the worst famine year, as the 1995 historical floods began to devastate large areas of the country. In this light, the finding that the very young cohorts born before the peak of the famine are worse-off compared to those born afterwards seems to make sense: they were exposed to the worst femine in1995 thus, they simply could not recover thereafter. Whereas cohorts that avoided this year of the extreme crisis, seem to be relatively taller. We control for location, where we limit ourselves to the province level. Location indicates the child’s current residence, and in the totalitarian North Korea, current residence is likely equal to birthplace, as the regime strictly regulates internal migration for political reasons. This way, one can also control for differences in regional living standards at birth. As a result, administrative provinces come out statistically significant as well: Compared to South Pyongan province, a main mining region in North Korea, cohorts born elsewhere are better-off, as clearly indicated by the positive (significant) province dummy coefficients in Table 3, regressions 2, 3, and 4. Coal is North Korea’s most abundant mineral resource. In the Cold-War era, South Pyongan was the main coal mining district (Savada, 1994). Yet, as discussed above, with the disintegration of the Eastern Bloc, North Korea’s heavy-industrial sector completely broke down due to lack of energy and mechanical spare parts. For this reason, South Pyongan seems to be worse-off than the rest of the country, as standards of living have deteriorated drastically compared to more agricultural regions. When comparing the seasonal effects found in regressions 1 and 2, birth year, sex, and location do not affect the results found in the basic model. Although these

ARTICLE IN PRESS D. Schwekendiek / HOMO — Journal of Comparative Human Biology 60 (2009) 59–75

69

Table 3. Linear regressions of HAZ on birth seasons.

(Constant)

HAZ

HAZ

HAZ 1997 survey HAZ 2002 survey

(1)

(2)

(3)

Coeff.

p

1.67

0.00 1.95

Birth season Spring (Mar–May) 0.08 Summer (Jun–Aug), ref. Autumn (Sep–Nov) 0.06 Winter (Dec–Feb) 0.08 Sex (male ¼ 1) Birth year 1991 1992 1993 1994 1995, ref. 1996 1997 1998 1999 2000 2001 2002 Birth province Pyongyang Nampo Kaesong South Pyongan , ref. North Pyongan South Hwanghae North Hwanghae Kangwon South Hamgyong North Hamgyong Ryanggang F Adj. R2 N

p

Coeff.

p

0.00 1.97

0.00

2.12

0.00

0.01 0.14

0.00 0.36

0.00

0.02

0.58

0.08 0.11 0.02 0.11 0.09

0.00 0.18 0.00 0.35 0.00 0.10

0.00 0.00 0.01

0.09 0.01 0.07

0.02 0.74 0.00

0.12 0.25 0.15 0.05

0.04 0.17 0.00 0.21 0.05 0.17 0.44 0.10

0.02 0.01 0.04 0.23

0.29 0.19 0.13 0.12 0.10 0.07 0.67

0.00 0.01 0.06 0.07 0.13 0.26 0.00

0.00 0.00

0.22 0.18 0.17 0.17 0.20 0.38 0.93

0.30 0.40 0.42 0.41 0.35 0.07 0.00

0.45 0.58 0.15

0.00 0.00 0.01

0.30 0.43 0.01

0.00 0.00 0.86

0.17 0.43 0.28 0.50 0.28 0.21 0.04

0.00 0.00 0.00 0.00 0.00 0.00 0.45

0.01 0.09 0.19

0.87 0.11 0.00

0.18 0.07 0.18

0.00 0.18 0.00

36.69 0.1065 5992

0.00

Coeff.

p

(4)

Coeff.

0.76 1.99

0.79 0.21 0.54 0.57

8.35 0.00 30.93 0.00 43.85 0.0022 0.0701 0.1321 9934 9934 3942

Shadowed areas: significance at the 5% level, ref. ¼ reference category.

0.00 0.04 0.00 0.00

0.00

ARTICLE IN PRESS 70

D. Schwekendiek / HOMO — Journal of Comparative Human Biology 60 (2009) 59–75

Table 4. Descriptive statistics. N

Percent

Cumulative percent

Season of measurement Summer (Aug) Autumn (Sep–Oct) Total

3827 6107 9934

38.50 61.50 100

38.50 100.00

Season of birth Spring (Mar–May) Summer (Jun–Aug) Autumn (Sep–Nov) Winter (Dec–Feb) Total

2577 2457 2304 2596 9934

25.94 24.73 23.19 26.13 100

25.94 50.67 73.87 100

Year of birth 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Total

1230 755 481 598 506 700 532 621 723 1160 1603 1025 9934

12.38 7.60 4.84 6.02 5.09 7.05 5.36 6.25 7.28 11.68 16.14 10.32 100

12.38 19.98 24.82 30.84 35.94 42.98 48.34 54.59 61.87 73.55 89.68 100

Province Pyongjang Nampo Kaesong South Pyongan North Pyongan South Hwanghae North Hwanghae Kangwon South Hamgyong North Hamgyong Ryanggang Total

598 599 600 1624 600 1431 799 853 1631 599 600 9934

6.02 6.03 6.04 16.35 6.04 14.41 8.04 8.59 16.42 6.03 6.04 100

6.02 12.05 18.09 34.44 40.48 54.88 62.93 71.51 87.93 93.96 100

ARTICLE IN PRESS D. Schwekendiek / HOMO — Journal of Comparative Human Biology 60 (2009) 59–75

71

controls considerably increase the adjusted R squared to 7%. In fact, they even corroborate the seasonal effects, as the autumn dummy now clearly turns significant at the 1% level. As distinct from the pooled regressions 1 and 2, we conducted a separate analysis for the 1997 survey (regression 3), and for the 2002 survey (regression 4). As a main result, we can see a positive and statistically significant autumn effect in both surveys. However, for the 2002 survey, the negative winter and spring effects do not come out statistically significant, though with the same and expected sign as before. As an explanation of this, it may be argued that the nutritional situation had changed in 2002: by that time, food security throughout the year has slightly improved due to a steady flow of international food aid. It has been shown earlier that United Nations aid had a positive impact on the well-being of children at that time (Hoffman and Lee, 2005; Schwekendiek, 2007). For this reason, there might be no statistically significant effect on the winter and spring cohorts. The situation might have differed from the 1997 famine peak, when there was no or very little food available throughout winter until spring (the commonly called ‘‘lean’’ season, besides famine). Because there is still a statistically significant autumn effect in 2002—thus, after the peak of the famine—it may be tentatively concluded that the harvest season might have an impact on the development of heights. Finally, both surveys were carried out in early autumn (Table 4). It should be noted that the date of measurement could have biased the results: rates of human growth in stature differ by months of the year, so seasonal fluctuations in height outcomes could partially depend on the timing of the measurement. Yet, we have used z-score transformation where the age of individuals in the study group and of the corresponding reference group was measured in months, thus, these distortions should play a minor role.

General discussion There are potential shortcomings of taking the birth season as a determinant of stature. Five issues will be addressed here: the spurious correlation hypothesis, the survivor bias hypothesis, the workload hypothesis, concerns over eating habits, and the statistical artefact hypothesis. The first four hypotheses are more or less particular manifestations of an omitted variable bias (OVB), whereas the fifth simply implies a general criticism against the statistical detection of seasonal birth patterns in assessing health status. First and foremost, the spurious correlation hypothesis implies that omitted social variables might matter for the development of stature. Depending on their social strata, parents might deliberately plan their child’s birth and have preferences for certain sections of the calendar. Thus, the birth month would be an indicator of the social background and education of the parents. This could again be mediated to their offspring’s biological welfare. Are birth month and social status and education correlated? Doblhammer (2004) found this effect for Austria using 1981 census data. According to her studies, educated parents seem to have a preference for spring

ARTICLE IN PRESS 72

D. Schwekendiek / HOMO — Journal of Comparative Human Biology 60 (2009) 59–75

births, whereas uneducated parents prefer autumn births. However, Doblhammer also points out that the birth month pattern hardly has any influence on health. As for socialist North Korea, height differences by education do exist, but they do not seem to be large. North Korean male and female refugees holding a college degree were found to be just 1.86 and 0.29 cm taller than those who only completed secondary education (Pak, 2004). This points towards moderate, not drastic,and differences in living standards by level of education. Lam and Miron (1991) noted, that birth month planning is driven by both social and biological factors. It may be doubted that social factors matter during a national famine which affects virtually all households. Thus, it may be assumed that child birth seems to be biologically rather than socially stimulated, and that social factors would not severely bias our findings. The survivor bias hypothesis claims that the survey measured the fittest individuals, whereas weaker individuals may have died during the food crisis. If selective survival was caused by seasonality, this could bias the seasonal birth patterns in heights. This can be checked by looking at the distribution of births by season. As discussed, there is no indication of social birth month planning in North Korea. Hence, we would expect this distribution to be roughly uniform. This seems to be the case, when looking at Table 4, where the absolute number of children born from 1991 to 2002 is shown by season. However, we see a minor drop in number of births for summer and autumn, which could indicate biological robustness. Yet as the distribution does not seem to be heavily biased, we can conclude that selective survival is supposed to play a minor role in explaining seasonal patterns in health. The workload hypothesis assumes that anthropometric seasonal effects could be caused by labour seasonality. In the case of a lasting period of heavy workload, physical growth is retarded, as nutritional inputs are first used for physical maintenance. Generally, it should be mentioned that the workload hypothesis does not become a direct problem for the analysis, because it exclusively takes into account infants and pre-school children for whom child labour is not an issue. However, labour seasonality could be mediated through working mothers. In particular, communist North Korea promotes the idea of working women. Heavy workloads might also matter given that the Stalinist regime opted for mass mobilisation instead of modernisation. The physical workload is, in general, relatively high for manual workers, specifically males in the North Korean labour ‘market’. It is generally low for women—thus for working mothers—who are commonly office workers, teachers, etc. in the North Korea. In addition to this, mass mobilisations are widely employed in agriculture and construction, and work intensity in these sectors is highly seasonal. The military is especially used for these mass mobilisations. We can reasonably assume that the mothers of the children examined in the sample are wives. Because primarily unmarried women serve in the army, we can conclude that these mass mobilisations should not affect our sample very much. For all these reasons, a seasonally varying workload is not likely to have caused the anthropometric seasonal effects we detected. A fourth frequently raised concern is that cultural eating habits—deeply rooted in religion—cause seasonal patterns in health. The Christian Lent or Muslim Ramadan might have a negative impact on the cohorts born in the corresponding periods, as

ARTICLE IN PRESS D. Schwekendiek / HOMO — Journal of Comparative Human Biology 60 (2009) 59–75

73

the nutritional supply for the foetuses will be lower at those times. This concern can be ruled out because we do not find these religious elements in North Korean society. The communist government oppresses all religious groups, and instead, worshipping the North Korean leader(s) has replaced religious elements. The government has not introduced any cultural eating habits nor promoted the idea of nutritional asceticism for certain periods of the year. It should also be noted that eating habits or cultural eating restrictions are often flexible for pregnant women, e.g. in Islam, women are permitted to avoid Ramadan in pregnancy if they want to. Last but not least, an overall concern raised by opponents of the seasonal birth concept is that seasonal effects are a mere statistical artefact. However, this paper has consistently identified statistical seasonal effects in two surveys—in 1997, which was at the peak of the North Korean food crisis, and in the post-famine period of 2002 (Table 3, regressions 3 and 4). Thus, in two cross-sections, we see a positive and statistically significant autumn effect on heights—which casts doubt on the existence of an artefact. As discussed, the negative winter effect is also consistent with other studies on stature, whereas we could not confirm the positive spring effect found by other researchers. Summarising; this paper primarily investigated birth season as an early-life determinant of the biological standard of living in North Korea during the food crisis of the 1990s. For this purpose, children’s heights as a sensitive indicator of the biological welfare of a nation were taken into account. Using regression analysis, we could verify a number of studies that found a negative effect for the winter cohorts. We did not detect the positive spring effect found in previous studies on birth season and stature, but we found a positive autumn effect. One possible explanation could be that the national harvest could have mattered since seasonal results in heights are correlated to the national crop calendar. Yet, as harvested food is not directly consumed in early-life, these findings require further investigation, as epidemiological and biological early-life factors have an impact on stature (Tanner, 1990; Doblhammer, 2004). When controlling for sex, birth year, and birthplace, these anthropometric seasonal effects remain robust. Reviewing some (though certainly not all) potential distortions, it has been concluded that social differences, a heavy workload, selective survival, and eating habits are not likely affect the results substantially. Furthermore, as these seasonal effects are consistent across two cross-sectional surveys, the paper provided evidence against a specific statistical artefact. These results may have practical implications. In general, for political decision makers, it might make sense to interfere in the birth calendar to avoid consequences of food shortages. Ignoring all humanitarian dilemmas, it can be done by efficient intervention in the birth calendar by giving governmental orders or launching propaganda campaigns, such as encouraging mothers to plan birth of the child in the early autumn season and by discouraging giving birth in winter. Obviously, in case of totalitarian North Korea, Pyongyang can easily influence the biological welfare of its people. In view of North Korea’s economic isolation as a consequence of its recent nuclear testing, this might be a feasible political option for the regime to reduce the magnitude of a potential famine.

ARTICLE IN PRESS 74

D. Schwekendiek / HOMO — Journal of Comparative Human Biology 60 (2009) 59–75

References Axt, P., Axt-Gadermann, M., 2004. Mai-Frau sucht Dezember-Mann. Herbig, Mu¨nchen. Banegas, J., Rodriguez-Artalejo, F., Graciani, A., de la Cruz, J., Gutierrez-Fisac, J., 2001. Month of birth and height of Spanish middle-aged men. Ann. Hum. Biol. 28, 15–20. Baten, J., Wagner, A., 2003. Autarchy, market disintegration, and health: The mortality and nutritional crisis in Nazi Germany 1933–1937. Econ. Hum. Biol. 1, 1–28. CBS. 2004. DPRK 2004 nutrition assessment. Report of survey results. CBS, UNICEF, WFP. 2002. Report on the DPRK nutrition assessment 2002. Centola, G., Eberly, S., 1999. Seasonal variations and age-related changes in human sperm count, motility, motion parameters, morphology, and white blood cell concentration. Fertil. Steril. 72, 803–808. Chenoweth, L., Canning, R., 1941. Relation of season of birth to certain attributes of students. Hum. Biol. 13, 533–540. Cole, T.J., 1993. Seasonal effect on physical growth and development. In: Ulijaszek, S., Strickland, S. (Eds.), Seasonality and Human Ecology. Cambridge University Press, Cambridge, pp. 89–106. De Onis, M., Blo¨ssner, M., 1997. WHO Global Database on Child Growth and Malnutrition. World Health Organization, Geneva. Doblhammer, G., 2004. The Late Life Legacy of Very Early Life, Berlin [et al.], Springer. Doblhammer, G., Vaupel, J., 2001. Lifespan depends on month of birth. In: Proc. Nat. Acad. Sci. USA 98, pp. 2934–2939. Ferry, J., Rolf, K., Troesken, W., 2006. The past as prologue. Paper presented at the Third International Conference on Economics and Human Biology, Strasbourg, 22–24 June. Fogel, R., 1983. Secular changes in American and British stature and nutrition. J. Interdiscip. Hist. 14, 445–481. Goodkind, D., 1999. Do parents prefer sons in North Korea? Stud. Fam. Plann. 30, 212–218. Henneberg, M., Louw, G.J., 1990. Height and weight differences among South African urban schoolchildren born in various months of the year. Am. J. Hum. Biol. 2, 227–233. Henneberg, M., Louw, G.J., 1993. Further studies on the month-of-birth effect on body size: rural schoolchildren and an animal model. Am. J. Phys. Anthropol. 91, 235–244. Hoffman, D., Lee, S-K., 2005. Prevalence of stunting and wasting in the Democratic People’s Republic of Korea. J. Nutr. 135, 452–456. Huntington, E., 1938. Season of Birth. Wiley, New York. Katona-Apte, J., Mokdad, A., 1998. Malnutrition of children in the Democratic People’s Republic of North Korea. J. Nutr. 128, 1315–1319. Komlos, J., Baten, J. (Eds.), 1998. The Biological Standard Of Living in Comparative Perspective. Steiner, Stuttgart. Lam, D., Miron, J.A., 1991. Seasonality of births in human populations. Soc. Biol. 38, 51–78. ´ Gra´da, C. (Eds.), Macintyre, K., 2002. Famine and the female mortality advantage. In: Dyson, T., O Famine Demography: Perspectives from Past and Present. Oxford University Press, Oxford. Mills, C., 1941. Mental and physical development as influenced by season of conception. Hum. Biol. 13, 378–389. Otto, W., Glaass, R., 1959. Inwieweit sind Tragezeit, Ko¨rperla¨nge und Ko¨rpergewicht der Neugeborenen von jahreszeitlichen Einflu¨ssen abha¨ngig? Deutsch. Gesundheitswes. 14, 1625–1628. Otto, W., Noack, H., 1957. Jahreszeitliche Schwankungen der Ko¨rpermasse bei 100000 neugeborenen Berliner und Leipziger Kindern. Zur menschlichen Vererbungs—Konstitutionslehre 34, 250–260. Pak, S., 2004. A study of North Korean biological standards of living using anthropometric data from North Korean escapees. Paper presented at the Second International Conference on Economics and Human Biology, Munich 2–6 June. Palka, E., Galgano, F., 2004. North Korea. McGraw-Hill, Guillford. Philips, D., Young, J., 2000. Birth weight, climate at birth and the risk of obesity in adult life. Int. J. Obes. Relat. Metab. Disord. 24, 281–287. Savada, A.M., 1994. North Korea. A Country Study. Library of Congress, Washington D.C. Schwekendiek, D., 2007. Health economics in the DPRK. Paper presented at the Fourth Augustin Cournot Doctoral Days. Strasbourg, 10–12 April. Schwekendiek, D., 2008. The North Korean standard of living during the famine. Soc. Sci. Med. 66, 596–608. Tanner, J., 1990. Foetus into Man. Harvard University Press, Cambridge. Torrey, E.F., Miller, J., Rawlings, R., Yolken, R.H., 1997. Seasonality of births in schizophrenia and bipolar disorders: a review of literature. Schizophr. Res. 28, 1–38.

ARTICLE IN PRESS D. Schwekendiek / HOMO — Journal of Comparative Human Biology 60 (2009) 59–75

75

Waterlow, J., Buzina, R., Keller, W., Lane, J., Nichaman, M., Tanner, J., 1977. The presentation and use of height and weight data for comparing the nutritional status of groups of children under the age of 10 years. Bull. WHO 55, 489–498. Weber, G., Prossinger, H., Seidler, H., 1998. Height depends on month of birth. Nature 391, 754–755. Wohlfahrt, J., Melbye, M., Christens, P., Andersen, A., Hjalgrim, H., 1998. Secular and seasonal variation of length and weight at birth. Lancet 52, 1990. Xu, X., Wang, W., Guo, Z., Cheung, Y., Karlberg, J., 2001. Seasonality of growth in Shanghai infants (n ¼ 4128) born in 11 consecutive years. Eur. J. Clin. Nutr. 55, 714–725.