During the demographic transition, fertility decline follows a period of decreasing mortality. While this transition is a universal phenomenon, its characteristics and speed of progression vary considerably across continents, or even between neighbouring countries. Using a rich dataset covering five countries of Central Asia – Kazakhstan, Kyrgyzstan, Tajikistan, Turkmenistan and Uzbekistan – Thomas Spoorenberg analyses fertility trends in a region long dominated by Russia, even before the Second World War. He shows that fertility initially increased before levelling off and then declining. He analyses the reasons behind this fertility pattern in each country, and the specific behaviours of certain sub-populations by sex and ethnicity. Proposing various hypotheses linking urbanization, improved health and economic development to the observed fertility trends, he shows how analysis of long-term patterns sheds light on the fertility transition.
1Central Asia (namely Kazakhstan, Kyrgyzstan, Tajikistan, Turkmenistan and Uzbekistan) is a unique laboratory for demography. The region includes countries with diverse social and economic profiles and offers singular opportunities to explore the effect of some major historical changes on the course of demography. The establishment of communism early in the twentieth century and the break-up of the Soviet Union in 1991 can be both considered as unique "natural social experiments". Yet, the demography of Central Asia remains poorly studied.
2Although the recent effect on fertility and family-building of the collapse of the Soviet Union, the resulting economic crisis and subsequent recovery have been well documented for Central Asia (Agadjanian et al., 2013; Barbieri et al., 1996; Dommaraju and Agadjanian, 2008; Spoorenberg, 2013, 2015a), the establishment of socialism in the early twentieth century in the region and its implications for the onset of the fertility transition has received very limited attention so far.
3Following the expansion and spread of the Russian Empire to Central Asia during the nineteenth century, Central Asian countries became part of the Soviet Union when it was founded in 1922. Their initial status was that of autonomous republics within the Russian Soviet Federative Socialist Republic (RSFSR), and it was only few years later that the Soviet Republics of Central Asia were carved out of the RSFSR. [1] The socialist model of development in Central Asia advanced markedly during the late 1920s and throughout the 1930s under Stalin's programme of forced sedentarization of nomads and collectivization of the economy. At the same time, traditional political circles were replaced by local leaders embracing the Stalinist ideology.
4The development of Central Asia under the Russian imperialist and Soviet rules can be seen as a “massive attempt to transmit the main elements of European culture to a non-European people” (Medlin et al., 1971, p. 7). The Russian imperialist and Soviet engagement in the region was pursued with the ostensible aim of bringing enlightenment and civilization to “backward” (nomadic) people of the steppes. The Soviet authorities emphasized the need for “inculcation of civilized behavior – hygienic habits, refined manners, and proper comportment” (Hoffmann, 2003, p. 17). The Sovietization and modernization of Central Asia in the early twentieth century resulted in the industrialization of the economy, substantial improvement in living standards, education and health conditions, and the secularization of the indigenous (Muslim) populations. [2] The magnitude of these developments has potentially strong implications for demography, and especially fertility.
5To the best of our knowledge only one study has examined in some detail the onset of the fertility transition in Central Asia. [3] Analysing the change in nuptiality and marital fertility in the Soviet republics using the Princeton indices (If, Ig and Im) [4] computed on data from the 1897, 1926, 1959 and 1970 population censuses, [5] Coale, Anderson and Härm (1979) have shown that marital fertility increased between 1926 and 1959 and 1959 and 1970 in all the Soviet republics of Central Asia. While providing valuable information on the changes in reproductive behaviours from the early twentieth century to the late 1960s in Central Asia, the Princeton indices cannot readily be used to derive a clear picture of levels and trends in fertility in the region that can be compared with other more recent studies. Moreover, the estimates of three single points in time separated by several years do not provide insights on what was happening between censuses.
6Mobilizing various data sources and estimation methods, this study reconstructs the fertility levels and trends of each Central Asian country from the early twentieth century in order to investigate the fertility response to the profound societal changes that were set in motion in the 1920s and 1930s and to propose an explanatory framework for the onset of the fertility transition in the region.
7As various studies have shown, social and economic development can influence fertility in two opposing directions in the early stage of the fertility transition. After a brief overview of the development in human capital, living standards, and health conditions in Soviet Central Asia over the first decades of the twentieth century, the article describes how changes in these areas are likely to influence the course of fertility. The study then presents the data and methods employed to reconstruct the fertility levels and trends in the region, focusing mostly on the 1930-1970 period. Despite different levels, the five countries display a similar pattern of fertility change. However, a closer analysis by main nationality/ethnic group in each Central Asian country reveals that the fertility rise was circumscribed to the indigenous ethnic groups only, with the fertility of the women of European origin remaining stable. The last section discusses the results presented and draws their implications for the study of the onset of the fertility transition globally.
I. Early twentieth century development in Central Asia
8Central Asian societies experienced rapid progress in the early years of the Soviet period. Women’s occupations, and their status and role in society changed considerably (Ishkanian, 2003. Lubin, 1981). The impressive achievements in human capital and health helped to bring Central Asian women closer to their European Soviet counterparts. Perhaps the progress in human capital formation illustrates best the extent of these achievements; many were unprecedented in history. As a result of major investment in free and accessible education systems, female literacy (measured here as the proportion of people age 20 and over who completed at least primary education) rose spectacularly from around 25% in the early 1920s to 80% or above just 20 years later in the early 1940s (Figure 1). From fundamentally illiterate populations, the Central Asian countries became literate societies within two decades or so.
Figure 1. Proportions of the male and female population aged over 20 who completed at least primary education, Central Asia, 1900-1950
Figure 1. Proportions of the male and female population aged over 20 who completed at least primary education, Central Asia, 1900-1950
9Progress in human capital formation accompanied the urbanization process in the region. Despite changes in definitions of "urban" and "rural" areas between censuses, [6] the existing data point to a clear and increasing concentration of people in urban agglomerations, mostly due to the forced sedentarization of nomadic groups, population resettlement, and increased opportunities in urban areas following the industrialization of the economy. Figure 2 indicates impressive gains in urban population in each country of the region from the late 1920s to the late 1950s. In Turkmenistan, the urban population increased almost 2.5-fold between 1926 and 1939; in Tajikistan it nearly doubled between 1939 and 1959, and similar substantial increases were observed in other countries of the region. Between 1939 and 1959, while the total population of Central Asia increased by a factor of 1.4, its urban population was multiplied by 2.2.
Figure 2. Percentage of urban population, Central Asia, 1926, 1939, 1959 censuses
Figure 2. Percentage of urban population, Central Asia, 1926, 1939, 1959 censuses
10The urbanization of the populations of Central Asia gave rise to substantial investments in social and health infrastructures, including the creation of a public health system (Loring, 2008; Michaels, 2003), which brought undoubted improvements in living standards and health conditions. [7] Unfortunately, very few estimates are available to describe the health conditions in Central Asia in the early twentieth century. With respect to Soviet standards, the Central Asian republics had the worst health conditions in the Soviet Union (Lorimer 1946). Yet, there is no doubt that the countries of the region benefited from substantive health improvements in the first decades following the instauration of the socialist agenda (Wheatcroft, 1999). Although the official figures on infant mortality available for the Central Asian republics are notorious for severely underestimating mortality levels (Anderson and Silver, 1986; Guillot, 2007), official figures indicate that improvement was under way in the 1940s and beyond (Michaels, 2003). While life expectancy at birth was around 30-35 years in Central Asia in the mid-1920s according to independent estimates (Coale et al., 1979), it had increased to between 50-55 years by the early 1950s (United Nations, 2015). These figures indicate that the health transition began during the 1930s and 1940s in the region.
11The societies and economies of Central Asia experienced profound changes during the three decades between the late 1920s and the late 1950s. As we shall see, they had profound consequences for the demographic development of the region, fertility especially.
II. Conceptual framework
12In the classical formulation of the demographic transition, social and economic development is a major force leading to mortality reduction, followed later by fertility decline (Chesnais, 1986; Dyson, 2010). With regard to fertility, it is widely accepted that social and economic development (such as improvements in living standards and health status, better education, urbanization, etc.) gives rise to new social trends that encourage the adoption and spread of fertility control. While it is rarely posited that social and economic development can contribute to increased fertility by unleashing women’s childbearing potential through the reduction or removal of various biological and behavioural fertility checks, there is abundant evidence to support such a view (Dyson and Moore, 1985; Garenne, 2008; Ortega, 2009; Reher and Requena, 2014; Romaniuk, 1980, 1981; Saito, 2006).
13Among other variables, development can increase fertility through changes in breastfeeding, nutrition, fecundity (effects of venereal diseases), mortality/survival, age at marriage and proportion never-married (Nag, 1980). Breastfeeding can prolong the period of postpartum amenorrhea, and thus affect fertility. Development, through rural migration to urban areas, can reduce the practice of breastfeeding and its duration, thereby increasing fertility by shortening the period of postpartum amenorrhea and the interval between births (Knodel, 1977). In addition, female fecundity is assumed to increase with development, because women's nutritional status usually improves as a consequence and because well-nourished lactating mothers have shorter periods of amenorrhoea (Frish, 1975). Indeed, it has been estimated that the abandonment of breastfeeding can contribute to a reduction in average birth intervals of between 14% in a well-nourished population and 40% in a population where nutrition is barely adequate, increasing the level of fertility by 16% and 64%, respectively (Knodel, 1977, based on Potter, 1975).
14Beside breastfeeding and nutrition, the prevalence of venereal diseases (syphilis, gonorrhoea, etc.) is a major involuntary factor contributing to fertility reduction (Belsey, 1976; Gray, 1979; McFalls and McFalls, 1984). Because advances in medicine and public health can reduce the prevalence of venereal diseases, development is assumed, therefore, to raise the level of fertility where these diseases are prevalent.
15In the classical theory of the demographic transition, mortality decline precedes fertility decline. Lower mortality can, however, contribute to increased fertility by reducing the proportion of widows/widowers, thereby extending the duration of marriage and hence the time a woman is exposed to reproduction (Nag, 1980).
16Lastly, changes in age at marriage and in the proportion never-marrying (especially among women) are two of the most influential variables for fertility (Bongaarts, 1978). Development is usually associated with a postponement of marriage and an increase in the proportion of women who remain single, thus contributing to reduced fertility. However, evidence from less developed countries indicates a decline in the age at marriage and the proportion of never-married women over the course of development, with a positive effect on fertility levels (United Nations, 1977).
17The positive or negative roles of development on fertility are indeed not antithetical and can be reconciled in a common framework. The two series of factors can influence fertility sequentially and result in a temporary increase in fertility when the forces favouring the adoption and spread of fertility control are too weak to counterbalance those unleashing women’s childbearing potential (Spoorenberg, 2015b). Under this sequential perspective, the fertility rise that precedes the continued fertility decline should therefore be considered as being one possible phase in the onset of the fertility transition (Dyson and Murphy, 1985).
18A recent reconstruction of the historical fertility change in Mongolia – also a former socialist country – shows that the improvements in health status and living standards linked to the adoption and development of socialism led initially to a substantial fertility increase of about 2.5 children per woman (mainly through the reduction of disease-related sterility and the processes of sedentarization and urbanization) that was followed two decades later by a prolonged fertility decline (Spoorenberg, 2015b). Given the historical and developmental similarities between Central Asian republics and Mongolia, it is likely that similar mechanisms drove fertility trajectories in Central Asia in the early Soviet decades.
III. Data and methods
19As part of the state apparatus, the Soviet Republics of Central Asia organized the collection of statistical data in order to plan and monitor the socialist model of development. Yet, very few studies have taken advantage of these data to study population developments in the countries of the region. This study is based on a large series of estimates of total fertility obtained by applying direct and indirect estimation methods to population statistics available from the civil and vital registration system, population and housing censuses, and household sample surveys.
20Population statistics from the 1926, 1939, 1959, 1970, 1979, and 1989 population and housing censuses conducted in the Soviet Union were used to reconstruct fertility changes in each country of region and for the main ethnic groups within a country. Fertility estimates from sample surveys and civil and vital registration system data were also included for comparison purposes.
21As few direct estimates are available, the changes were estimated mostly using indirect demographic techniques. Two demographic methods were used to estimate and reconstruct the fertility levels and trends in the countries of the region, namely the reverse survival method of fertility estimation and cohort fertility based on the number of children ever born classified by the mother's age. As we shall see, the two methods yield consistent fertility estimates that reconstruct fertility changes over almost a century.
22The reverse survival method is used to estimate fertility based on a population distribution by single age and sex collected during a census (or survey). Using a set of age-specific mortality and fertility patterns, the single-age population under age 15 is used to approximate the number of births in the past and derive a series of total fertility rates (Timæus and Moultrie, 2013). Population data by single age and sex are available from different census volumes that can be accessed in Demoscope Weekly, a newsletter published in Russian by the Institute of Demography at the National Research University Higher School of Economics in Moscow. The Excel template “FE_reverse_9.xlsx” provided with Timæus and Moultrie (2013) was used to estimate total fertility. The required mortality (5q0 and 45q15) and fertility estimates were taken from the World Population Prospects: The 2015 Revision (United Nations Population Division, 2015) and the Coale and Demeny Model West was used in the estimation procedure. The quality of the fertility estimates produced by the reverse survival method depends mostly on the quality of age declaration and on the effect of international migration (Spoorenberg, 2014). Given the importance of documentation in socialist regimes, age data is generally of good quality in population censuses conducted during Soviet times. Furthermore, the control exercised over the population by the Soviet authorities limited the scale of international migration. For these reasons, it should be possible to obtain consistent fertility estimates by applying the reverse survival method of fertility estimation to the population censuses conducted during and after Soviet times.
23The second indirect method uses the number of children ever born (CEB) by age of the mother. The total fertility at the time a cohort reaches its mean age at childbearing can be approximated from the cohort fertility using Ryder’s correspondence between period and cohort measures (Ryder, 1964, 1983). A mean age at childbearing of 28 years was used to determine the reference date of the cohort fertility (for details of time translation, see Feeney, 1995, 1996, 2014). CEB data are not available for all censuses and countries; Kazakhstan, Kyrgyzstan and Tajikistan have the largest amount of available data. Census samples are available in IPUMS-International (Minnesota Population Center, 2015) for the 1999 and 2009 censuses of Kyrgyzstan, and children ever born by single year of age were computed based on declarations of women aged 45-90. The CEB data for the 1979 census were made available by Dmitriy Bogoyavlenskiy from the Institute of Demography, National Research University Higher School of Economics in Moscow. The 1989 CEB data are available in Demoscope Weekly.
1. Quality of fertility estimates
24The fertility increase in Central Asia in the 1950s and 1960s is often attributed to improvements in the registration of births (Anichkin and Vishnevsky, 1994; Blum, 1987). However, while birth registration has certainly improved over the decades, this explanation is not verified by changes in the population structure observed in successive population censuses. A look at the population pyramids given by the 1959, 1970 or 1979 population census indicates a clear widening of their base starting in the early 1950s (not shown). Such a trend is indeed consistent with a pre-decline fertility increase. Further, because census data are usually collected from household heads, they would not be affected by any change in the quality of the vital registration system. Other factors may ultimately affect the quality of the fertility estimates derived from census data.
25Different data quality patterns may affect the population age-sex structures collected in population and housing censuses. Populations at select ages are often under-enumerated and age declaration is affected by heaping on selective age digits, thus affecting the fertility estimates based on a population age structure. Typically, young children are under-represented in population censuses. When fertility is estimated using the population age-sex structure, this may lead ultimately to an under-estimation of fertility in the years preceding the census. Children’s age misstatement results in a transfer to older ages and an attraction for ages 5, 10 and 15. While the transfer to older ages leads to further under-estimation of fertility in the years preceding the census, the heaping on ages 5 and 10 contributes to over-estimation of fertility around the fifth and tenth years before the census. Finally, the level of fertility would likely be under-estimated at the end of the 14-year period before the census due to an attraction for age 15. Because the quality of the age declaration of women age 10-64 also affects the estimation of fertility, it is not always straightforward to single out different biases in the reverse survival fertility estimates. In this study, the census age distributions used to compute the reverse survival fertility estimates were not corrected for these biases.
26Mortality is another factor liable to affect the quality of reverse survival fertility estimates. If the selected mortality level is too low, too few births will be added in the population, contributing ultimately to an under-estimation of past fertility levels. As described above, Coale-Demeny model life tables were used. While a given family of model life tables may not correspond to the mortality age patterns of the country studied, the reverse survival fertility estimates are only marginally affected by wrong selection of the level and age patterns of mortality (Spoorenberg, 2014).
27The data on the number of children ever born (CEB) are usually affected by recall lapse (women tend to omit children born many years earlier) (United Nations, 2004). Women may also omit to declare deceased children or children living elsewhere. In addition, information on CEB can only be collected among mothers who are present at the time of the census. One cannot rule out the possibility that these surviving women have fewer children on average, because the women who may have had more children were subjected to higher mortality risks delivering higher-order births and are therefore not present in the population during the census. The levels of total fertility based on cohort fertility data are therefore likely to be under-estimated.
28Finally, the fertility estimates based on reverse survival and cohort fertility can be affected by migration. Fertility may be over-estimated (under-estimated) as a result of the departure (arrival) of women and girls. For migration to affect fertility, female migrants must have either left their children behind (thus affecting only the denominator) or had a significantly higher or lower number of children (thus affecting only the numerator). Similarly, cohort fertility can potentially be affected by migration if CEB works as a selection mechanism to migration. As a final point, for either method, only the migration of large portions of the female population would affect the estimation of fertility. In the case of Central Asian countries, large immigration flows of European-origin populations occurred during the twentieth century (Rahmonova-Schwarz, 2010). These arrivals could affect the estimation of fertility because these migrants mostly originated from areas/republics where fertility was already lower than in Central Asia. These inflows would therefore contribute to lower fertility at national level.
IV. Fertility levels and trends in Central Asia
29The fertility estimates obtained by applying the different estimation methods to the available data are presented for each country in Figure 3. They give a fairly consistent picture of fertility change over almost a century in the region. In each country, total fertility fluctuated around 4-5 children per woman in the decades before 1940, with the exception of Kazakhstan where fertility dropped markedly in the early 1930s as a consequence of famine. Fertility started to increase around the early 1940s then reached a plateau in the late 1950s and throughout the 1960s. By the mid-1970s, fertility had begun a prolonged decline that was momentarily reversed or slowed down in the late 1980s following the introduction of new welfare benefits (labour law on maternity and childcare leave, social benefits for children, etc.) (Jones and Grupp, 1987; Zakharov, 2008). Fertility bottomed out at the turn of the twenty-first century and subsequently levelled off or increased in all the Central Asian republics (Spoorenberg, 2015a).
Figure 3. Country-level estimates of total fertility between 1910 and 2010, Central Asia
Figure 3. Country-level estimates of total fertility between 1910 and 2010, Central Asia
30The fertility levels and trends in the 1910s and 1920s, based on the 1926 census, should be taken with caution because less importance was attached to recording age than in later Soviet times (and no attempt was made to correct the census age distribution for biases). Similar care should be exercised with the fertility estimates during the late 1920s and 1930s based on the 1939 census. The tumultuous period of the 1930s influenced the quality of the census taking and reduced fertility (especially in Kazakhstan where the effect of the famine is evident in the early 1930s).
31In all countries, fertility seems to have fluctuated widely in the 1910s, 1920s and 1930s. These variations could be related to the massive recruitment of soldiers from Central Asia during the First World War, the Basmachi rebellion that started to unfold in 1916, the start of the collectivization process in 1928 and, for Kazakhstan, the famine that struck in the early 1930s. More studies are needed to better understand the possible effects of these events on fertility levels and trends in the region.
32Numerous differences are observed between the reverse survival fertility estimates and the cohort estimates. While both series corroborate the fertility increase starting in the late 1940s, the increase suggested by the cohort fertility estimates is smaller (or almost non-existent in the case of Kazakhstan). Due to recall lapse, omissions of deceased children, and/or mortality selection (see above), the estimates based on cohort fertility are likely to under-estimate fertility.
33Despite variations due to the quality of the declaration of children’s ages (see above), the fertility estimates based on the reverse survival method are thought to be of better quality. They indicate a larger fertility increase in Tajikistan, Turkmenistan and Uzbekistan, followed by Kyrgyzstan and lastly Kazakhstan. Like today, the cross-country difference in the fertility increase is linked to the national population composition of each republic. The populations of Kazakhstan and Kyrgyzstan included a larger share of people of European origin. Already in 1939, almost 53% of the population of Kazakhstan was of Russian, Ukrainian, Belarussian or German origin, and in Kyrgyzstan close to a third of the population was of European origin. In Tajikistan, Turkmenistan and Uzbekistan, on the other hand, the share was much smaller, at between 10% and 20%. The differing population composition of each republic helps to explain the differential increases in total fertility at national level.
V. Ethnic divide in fertility
34All Central Asian countries shared a similar pattern of historical fertility change over the twentieth century: an early increase followed by a plateau and then a prolonged decline. Yet, this general pattern conceals interesting fertility trends within national populations. Indeed, the pre-decline fertility rise recorded across the region was specific to certain ethnic groups only.
35Estimation procedures similar to those used at the national level were applied to population censuses and surveys to derive ethnic-specific estimates of total fertility. Figure 4 shows the reconstruction of ethnic-specific fertility in each of the five Central Asian countries. Depending on data availability, the fertility estimates for the titular ethnic group, the European-origin group, and one of the other main indigenous ethnic groups are shown.
36Despite the differing number of data sources available for each country, there is a striking contrast between the fertility levels and trends of the titular and indigenous ethnic groups (i.e. Kazakh, Kyrgyz, Tajik, Turkmen, and Uzbek) on the one hand, and the European-origin group (mostly Russian) on the other.
37In fact, the fertility increase recorded at the national level (Figure 3) only concerned specific portions of the population. The increase between 1940 and 1970 occurred only among women of the titular and indigenous ethnic groups, while the fertility of women of European origin remained stable or declined steadily over the same period.
38Among the indigenous ethnic groups, the fertility increase was the largest in Kyrgyzstan and Turkmenistan, where more than two children were added to the fertility level. In Kazakhstan, Tajikistan, and Uzbekistan, the fertility of indigenous women increased by about two children per woman on average. The fertility series for the Russian ethnic group indicate much lower fertility levels and no fertility increase between 1940 and 1970. The Russian fertility trajectory in Central Asia mirrors the patterns observed in Russia during the same period (Avdeev and Monnier, 1995; Scherbov and Van Vianen, 2001).
Figure 4. Ethnic-specific estimates of total fertility (children per woman), Central Asia, 1910-2010
Figure 4. Ethnic-specific estimates of total fertility (children per woman), Central Asia, 1910-2010
39When the fertility trends and levels of Figure 4 are matched against the national ethnic population composition, fertility trajectories at national level can be better understood. Given that people of European origin, whose fertility did not increase between 1940 and 1970, made up more than half of the population of Kazakhstan, its fertility increase was much smaller than in the other countries of the region (see Figure 3). Furthermore, the fertility trajectory of the ethnic groups in each country may have been influenced by their size and relative importance, as well as their social position (Nedoluzkho, 2012).
40Figure 4 also presents some discrepancies between estimation series, especially between different estimates of Russian fertility when the series overlap in the late 1970s. For the 1989 census, no data are available for the population by single age and sex, so the estimates are based on 5-year age group data using the Sprague smoothing procedure. The difference between fertility estimates for Russia in the late 1970s is therefore mostly an artefact of the methods used to derive reverse survival fertility estimates from the available information. The change in the ethnic composition of the population could also be a factor. It is possible that the women of European origin who decided to leave the region after the 1979 census had fewer children on average than those of the same origin who stayed. Examining this assumption is beyond the scope of this article, however.
VI. Factors contributing to the fertility increase
41In order to better understand the fertility increase in Central Asia, a series of proximate factors of fertility were examined more closely. It is perhaps even more important to identify the factors that drove fertility because the increase recorded between 1940 and 1970 was indeed in contradiction with the expected direction of change in response to the industrialization and modernization of the early Soviet decades. With the development of (female) education, coupled with increasing female labour force participation and the concentration of the population in urban areas, one might have assumed that fertility would decline rather increase in such an impressive manner. We therefore investigated changes in marriage patterns, fecundity and sterility, breastfeeding and birth intervals, and in policy measures to determine the contribution of different factors to this fertility increase in Central Asia in the 1940s.
1. Change in the proportion of married women
42In societies where childbearing occurs within union, a change in marital behaviours is one of the main proximate determinants affecting the level of fertility (Bongaarts 1978). A surge in marriage could explain the observed fertility increase in Central Asia. To investigate this supposition, the proportions of married women by age in the successive Soviet censuses were used to examine changes in both the tempo and the quantum of marriage. Whereas the proportion of single women by age is usually considered when studying changes in marital behaviours, the available information on marital status by age in the 1939, 1959 and 1970 population censuses refers to married women only. The proportion of married women at age 20-24 was used to measure the change in the timing (tempo) of marriage, whereas the proportion of married women at age 35-39 served to assess the change in the intensity (quantum) of marriage.
43The proportion of married women at ages 20-24 in the successive Soviet censuses reveals that women in Central Asia postponed their marriage in the 1940s and 1950s (Figure 5A). The proportion of married women aged 20-24 declined sharply during the 1940s and 1950s, in Kazakhstan especially. While 78% of women in that country were married by age 20-24 in 1939, two decades later, in 1959, the proportion was just 62%. In the other Central Asian countries, despite less impressive declines, similar trends were observed. Female enrolment in education and the increasing involvement of women in the Soviet economy postponed entry into marriage in the region. During the 1960s, age at marriage levelled off and the proportions of married women at age 20-24 declined only slightly in the region, with the exception of Uzbekistan where a marked decline is recorded in the 1970 census.
Figure 5. Percentage of married women age 20-24 (A) and age 35-39 (B), 1939, 1959 and 1970 censuses, Central Asia
Figure 5. Percentage of married women age 20-24 (A) and age 35-39 (B), 1939, 1959 and 1970 censuses, Central Asia
44In terms of the intensity of marriage (measured here by the percentage of married women at age 35-39) (Figure 5 B), a slight decline was recorded between 1939 and 1959. This was followed by an increase during the 1960s in all countries of the region, except in Uzbekistan.
45Figure 5 shows that while fertility was increasing significantly during the 1940s and 1950s, marriage was being postponed and smaller proportions of women were getting married. A surge in marriage cannot, therefore, explain the fertility increase that began in the 1940s in Central Asia. Alternative factors need to be considered.
2. Fecundity and sterility
46Sexually transmitted diseases affect the chances of bringing to term and successfully delivering a pregnancy (Belsey, 1976; Gray, 1979; McFalls and McFalls, 1984). For example, among mothers with untreated syphilis, 40% of fetuses die in utero (Goldenberg et al., 2010).
47No definitive figure is available for the prevalence of syphilis and other sexually transmitted diseases during the early decades of the twentieth century in Central Asia, but there is no doubt that it was particularly high in the region (Clemow, 1903; Journal of The Royal Central Asian Society, 1959; Lorimer, 1946; Michaels 2003). Indeed, various expeditions to Central Asia reported prevalence rates of between 40% and 100% depending on the villages and regions surveyed. More complete health surveys conducted in 1928 indicated that 35% of the Kyrgyz population had syphilis (Loring, 2008).
48The development of a public health system (medical infrastructures and services) from the late 1920s in Central Asia, coupled with a change in diet and the availability of new medicines, may have reduced the prevalence of sexually transmitted infections, thus influencing women’s ability to achieve live births and thereby increasing the proportion of women having children. For example, the development of the public health infrastructure in Kazakhstan increased the number of women who sought prenatal care – during which they were screened for syphilis, among other conditions – and who received medical assistance during delivery (Michaels, 2003).
49The reported high prevalence rates of sexually transmitted diseases in the region are consistent with the fertility levels of 4-5 children estimated in the early twentieth century (see Figure 3). The high percentage of women affected by syphilis, gonorrhoea or other sexually transmitted infections could have served as a biological check on fertility. In Kyrgyzstan, for example, “infertility and impotence were very common, and even women who could bear children never had more than three or four” (Loring, 2008). The Soviet investments in health infrastructure and personnel, the propagation and inculcation of hygienic practices and the increased availability of treatments in the 1930s and 1940s may therefore have therefore have reduced the prevalence of sexually transmitted diseases in Central Asia, thus unleashing women’s childbearing potential and pushing up fertility.
50To investigate whether the fertility increase could be related to a reduction of infertility and pregnancy wastage, childlessness among women who had reached the end of their reproductive life (age 45 and above) was used as an indicator of (primary) infertility. [8] In populations where no effective birth control exists – especially among women without children – a high percentage of childless women may be due to pathological sterility (Romaniuk, 1980).
51Figure 6 shows the percentage of childless women by birth cohort computed based on the All-Union 1979 and 1989 censuses. In all five Central Asian countries, the percentage of childless women has declined both at the national level and among the ethnic groups. Interestingly, infertility fell more sharply among women of the indigenous ethnic groups than among women of Russian origin, even though a higher percentage of Russian women did not bear a child. The fact that the reduction of childlessness among Russian women in Central Asia was not accompanied by a fertility increase suggests distinct reproductive strategies and/or regimes between ethnic groups.
Figure 6. Percentage of childless women by birth cohort, Central Asia, 1979 and 1989 censuses
Figure 6. Percentage of childless women by birth cohort, Central Asia, 1979 and 1989 censuses
52The timing of the reduction of childlessness among the successive birth cohorts is consistent with the onset of the fertility increase 25-30 years later in the 1940s. As the proportion of women marrying did not increase during that period, the fertility increase could be explained by an increase in women’s fecundity.
53A decline in infertility would have affected the fecundity not only of nulliparous women, but of women at other parities as well. In order to measure the contribution of each birth to the fertility increase, parity progression ratios (PPRs) were computed based on the information on the number of children ever born by age of the mother collected in the 1979 census (Figure 7). [9] Only the PPRs of the titular ethnic group in each country are analysed here, because the fertility increase in each country occurred among indigenous ethnic groups (Figure 4).
54In all five countries, the pattern of fertility increase is largely similar (Figure 7). The reduction in childlessness contributed to the fertility increase by between 4% (Tajik women in Tajikistan) and around 10% (Kazakh women in Kazakhstan), and the largest contribution came from third and higher births. Although these changes are based on cohort fertility data that likely under-estimate the real patterns of change, the fertility increase in Central Asia is consistent with a reduction of both primary and secondary infertility in response to the development of the health system and the general increase in living standards throughout the region.
Figure 7. Percentage contribution of the change at each parity to the fertility increase among women of the titular ethnic group born in 1915-1919 and 1935-1939, Central Asia, 1979 census
Figure 7. Percentage contribution of the change at each parity to the fertility increase among women of the titular ethnic group born in 1915-1919 and 1935-1939, Central Asia, 1979 census
3. Breastfeeding and birth intervals
55The changes in the contribution of each birth to the fertility increase outlined in Figure 7 have implications in terms of birth intervals.
56Traditionally, breastfeeding was widely practiced among indigenous ethnic groups in Central Asia (Krieger 2006). [10] For fertility to increase as it did in Central Asia, major changes in breastfeeding practices must have taken place. Migration to urban areas is one of the developmental factors that may contribute to reducing the prevalence and duration of breastfeeding (Knodel 1977).
57To investigate the possible effect of urbanization on cultural practices related to breastfeeding and birth intervals, reverse survival fertility estimates were computed for urban and rural areas based on the 1959, 1970, 1979 and 1989 population censuses. An increase in the total fertility estimates in urban areas would provide an indirect confirmation of a possible negative effect of urbanization on breastfeeding.
58In each Central Asian country, the estimates of total fertility for urban areas increased by at least one child between the late 1940s and 1960s (not shown). Although fertility was likely increasing among women in urban areas, part of the urban fertility increase was indeed due to changes in the composition of the urban population following the arrival of women with higher birth rates from rural areas. The fertility increase was much larger in rural than in urban areas, so it is difficult to reach a firm conclusion regarding the effect of urbanization on breastfeeding in the region.
59Without data to substantiate any firm conclusions, one can conjecture that the increase in rural fertility was likely related to the collectivization of agriculture and population resettlement, as well as the general improvement in living standards, health conditions and medical infrastructures. Women living in rural areas experienced a larger fertility increase because the ongoing developments in agriculture, housing, living and health standards affected and benefited rural populations in particular. For example, in Kazakhstan and Kyrgyzstan, the forced collectivization and sedentarization of the nomadic population around urban clusters put an end to the traditional pastoral nomadic way of life. Such change is likely to have influenced the chances of achieving a live birth by reducing early miscarriages, thereby lowering the prevalence of childless women and shortening the interval between births. [11] In Tajikistan, the resettlement in the plains of populations from the foothills and mountains is also likely to have modified certain traditional ways of life, including breastfeeding practices.
4. Policy measures
60The Sovietization of Central Asia was accompanied by the replacement of the traditional political elites and the instauration of a new legal and policy environment. Among the policy measures that could have influenced the course of fertility, the Soviet parliament and government issued on June 27, 1936 a joint Decree on the Prohibition of Abortions, the Improvement of Material Aid to Women in Childbirth, the Establishment of State Assistance to Parents of Large Families, and the Extension of the Network of Lying-in Homes, Nursery schools and Kindergartens, the Tightening-up of Criminal Punishment for the Non-payment of Alimony, and on Certain Modifications in Divorce Legislation. Among other measures, the 1936 Decree allowed abortions under medical conditions only, imposed penalties upon persons performing abortions, and enacted a series of pronatalist measures to support women, families and children.
Ban on abortion
61Abortion is one of the basic proximate determinants of fertility identified by Bongaarts (1978). In the Soviet Union, abortion was decriminalized for the first time in 1920 and ever since has remained one of the most common forms of birth control in the region. The 1936 Decree could therefore have reduced the prevalence of abortions and contributed to the increase in fertility.
62From 1936 to 1955, abortion was made illegal, mainly as the result of Stalin’s concern about population growth in the Union. One might expect to observe a sharp increase in fertility just after the promulgation of the new law restricting abortion in 1936, as was the case in Romania the late 1960s (Teitelbaum 1972). Although an immediate increase in the number of births was indeed recorded in the year following the proclamation of the 1936 Decree, the effect was short-lived and the number of registered abortions continued to increase in the years after 1937 (Sakevich and Denisov 2014).
63Cohort fertility estimates for Kyrgyzstan – the only country of the region for which fertility estimates cover the 1930s – do not indicate an increase in the late 1930s following the ban on abortions in 1936. On the contrary, fertility seems to have dropped during that period. Indeed, it is likely that abortions continued to be practiced, though illegally (Michaels, 2003; Sakevich, 2015; Sakevich and Denisov, 2014), so the ban did not contribute to the fertility increase. Moreover, the women undergoing abortions were mostly of European origin. For example, data from an abortion clinic in Southern Kazakhstan indicate that only 5 out of the 1,997 abortions performed there were on Kazakh women (Michaels, 2003). Due to their lower fertility level, changes in the prevalence of abortion among women of European origin would only marginally influence the national fertility trajectory. Finally, although the indigenous population was only marginally concerned by legalized abortion, the social stigma associated with the procedure meant that women of indigenous origin wishing to terminate their pregnancy most likely resorted to unofficial practitioners and were therefore not counted in official statistics (ibid.).
Pronatalist policy measures
64Accompanying the ban on abortions in 1936, a series of policy measures were simultaneously adopted to support fertility and families, possibly contributing to the fertility increase. For example, mothers of large families (i.e. seven or more children, with the youngest child under age five) would receive 2,000 roubles per year for each child under age five, and women with eleven or more children (including one under age five) at the time of the decree in 1936 would receive 5,000 roubles in 1936 and 3,000 roubles per year thereafter (Michaels, 2003). Within less than three years, the number of kindergarten places tripled in the Soviet Union, rising from 700,000 to 2,100,000 by January 1939.
65Later on, as part of the efforts to stimulate population growth in the Soviet Union, a new decree to promote childbirth was issued in 1941 under which single and childless Soviet citizens were taxed. On 8 July 1944, the Soviet government promulgated a new Family Law that expanded social services available to mothers, particularly single women and mothers of large families. The state offered financial support to mothers based on the number of children they had. [12] The 1944 law was a response to wartime demographic losses and the hardships faced by women who were widowed or abandoned during the war (Michaels, 2003). Contrary to the ban on abortions that primarily affected women of European origin, the pronatalist policy measures mostly concerned indigenous women with higher fertility rates (Michaels, 2003). However, evidence of a positive effect of the pronatalist policy on fertility is very limited.
66The adoption of pronatalist policy measures, like the ban on abortions, likely played only a limited role in the fertility increase in Central Asia. While helping families to make ends meet, the family benefits were rather small in relation to the average Soviet wage (Heer and Bryden, 1966; Michaels, 2003). Furthermore, the experience of other countries where financial incentives were introduced to support childbearing shows that such incentives usually only affect the timing of fertility, rarely its intensity (Hoem, 1990; Frejka and Zakharov, 2013; Gauthier, 2007). In other words, women tend to adjust the timing of their childbearing in order to benefit from this support, but do not change the final number of children they have.
VII. Discussion and conclusion
67Despite its fascinating demographic history, Central Asia remains a world region poorly studied by demographers. The region is a unique demographic laboratory for studying the effects of historical change on the course of demography. The focus of this article was to investigate changes in fertility following the instauration of a socialist mode of development in the region in the early decades of the twentieth century.
68Fertility trends in Central Asia over the last hundred years follow an almost unique pattern, with similar fertility levels and breakpoints (Figure 8). In the early decades of the twentieth century, the number of children per woman fluctuated around 4-5 in the region (Period 1 and TFR 1). The rapid change and progress that unfolded in Central Asian societies from the 1930s were followed by an impressive rise in fertility across the region. Starting in the early 1940s, fertility increased steadily until the early 1960s in all the Central Asian countries (Period 2). After a plateau that lasted until the early 1970s (TFR 2), total fertility declined in all countries (Period 3) and reached a low point at the end of the twentieth century (TFR 3). Since then, fertility has been either stagnating or increasing in the region.
Figure 8. Stylized pattern of fertility change in Central Asia between 1910 and 2010
Figure 8. Stylized pattern of fertility change in Central Asia between 1910 and 2010
69Interestingly, the fertility increase in Central Asia was circumscribed to specific portions of the population; women of the titular and indigenous ethnic groups bore more children, but not women of European origin. This finding stresses the importance of looking beyond national-level aggregated indicators and points to the long-term coexistence of two distinct demographic regimes in each country of the region. The differences in demographic behaviours between ethnic groups observed today are the result of a durable legacy that has shaped the demographic development of the region for more than a century.
70From the late 1970s, the national fertility estimates presented in this study are generally consistent with the existing fertility series, confirming that the selected estimation methods produce consistent results. From the 1950s to the 1970s, however, some differences are found between the fertility estimates made by Blum (1987), the reverse survival fertility estimates, and the cohort fertility in Kyrgyzstan and Tajikistan. In general, Blum’s estimates are about one child lower than the other estimated series. Based on official Soviet statistics, Blum’s estimates use the annual number of registered births for some years in the period 1950-1973 [13] and the republics’ birth rates for the subsequent years. Blum’s series are likely to under-estimate the fertility levels in Central Asia because birth registration was incomplete in Central Asian republics for most of the years considered. Although some difference between estimation methods and data sources is to be expected, the comparison with the reverse survival fertility estimates suggest that birth registration in Central Asia became reliable in the late 1970s.
71The fertility estimates before the 1950s are consistent with earlier results based on the Princeton indices (Coale et al., 1979), but provide a fuller picture of ongoing fertility trends over several decades in Central Asia. Indeed, the application of "new" estimation methods to reconstruct long-term levels and trends in fertility at the national and ethnic levels makes it possible to revisit and expand the study of population development in regions of the world that have remained largely undocumented thus far.
72The examination of the possible contribution of the proximate determinants of fertility to the fertility increase in the first half of the twentieth century in Central Asia shows that the pre-decline fertility rise was not driven by a surge in marriage. In the 1940s and 1950s, women in Central Asia postponed their marriage and a higher proportion remained single. The data reviewed in this study suggest that an improvement in fecundity (and the decline in sterility) played a role in increasing fertility. Higher order births contributed the most to the fertility increase, possibly indicating a reduction in secondary infertility, as well as pointing to a reduction of birth intervals likely related, in the absence of contraception, to changes in breastfeeding practices. The social, economic and cultural changes brought by Soviet development seem to have contributed to the fertility increase by improving living conditions and health standards, and by modifying cultural and social practices of the indigenous populations of Central Asia. Although there is only indirect evidence to support these conjectures, the "routes" to increased fertility in Central Asia between 1940 and 1970 are consistent with the estimated effect on fertility of a change in breastfeeding practices linked to the nutritional status of the population. The fertility increase among the indigenous populations in Central Asia was lowest in Tajikistan – where fertility was already highest before the rise. Among the Tajik women in Tajikistan, fertility increased by almost 29%, only half the estimated maximum possible increase of 64% that can occur when breastfeeding is abandoned and nutrition is abundant (Knodel 1977).
73This study also points to several issues that deserve further consideration. First, efforts should focus on better estimating fertility levels and trends in the early twentieth century in the region, and on studying more closely the effect on fertility of the events that unfolded during that period (e.g. First World War, 1916 rebellion, and launch of collectivization). Such an endeavour would also need to consider the effect of policy changes associated with efforts to emancipate women, especially the legislative measures against early female marriage, forced marriage, kalym (bride price), and polygyny that were taken during this early period in Soviet Central Asia. Another research avenue would be to look at the political discourse on nation building, labour force shortage, etc. and their possible link to the fertility increase in the region.
74In more general terms, although it is widely accepted that social and economic development initiated forces that encourage the adoption and spread of fertility control, the case of the Soviet republics of Central Asia shows that development may also remove a series of biological and behavioural checks to women’s reproduction, thereby increasing fertility (Spoorenberg 2015b). The historical fertility development observed in the Central Asian republics adds yet another region to the growing list of countries where a pre-decline increase in fertility has been documented. The onset of the fertility transition in Central Asia suggests that when major advances in social and economic development occur rapidly, development influences fertility in two phases. First, when the organizational foundations of society are suddenly and profoundly redefined, development positively influences fertility because the changes are too great for women to adapt their childbearing behaviours immediately. Second, it is only later that the forces encouraging the adoption and diffusion of birth control are expected to reduce fertility. In this perspective, the fertility increase that precedes the continued fertility decline should therefore be regarded as one possible opening phase in the onset of the fertility transition (Dyson and Murphy, 1985). This conclusion therefore calls upon demographers to focus their attention on the multiple changes occurring during the opening phase of the fertility transition. A detailed examination of the factors driving the pre-decline fertility rise would allow population specialists to better understand the causes and timing of the subsequent fertility decline (ibid.).
75The study of the onset of the fertility transition in countries of Central Asia illustrates the need to reconstruct long-term demographic developments in developing countries in order to revisit their demographic transition in the light of the new availability of (old) datasets and/or the application of new estimation methods to existing population data. Such an endeavour will ultimately provide us with a broader and more accurate picture of the peculiarities of the global process of fertility transition.
The author wishes also to acknowledge the National Statistics Committee of the Kyrgyz Republic that provided some of the underlying data making this research possible.
Finally, the author thanks the three reviewers for their helpful com-ments and suggestions that contributed to improve the quality of this manuscript.
Notes
-
[*]
United Nations Population Division, New York, United States.
Correspondence: Thomas Spoorenberg, United Nations, Population Division/DESA, Population Estimates and Projections Section, 2 United Nations Plaza, Room DC2-1908, New York, NY 10017, United States, tel: +1 212 963 3214, email: spoorenberg@un.org -
[◊]
The views expressed in this article are those of the author and do not necessarily reflect the views of the United Nations.
-
[1]
The Turkmen Soviet Socialist Republic (SSR) and the Uzbek SSR were created in 1924; the Tajik SSR was an autonomous republic within the Uzbek SSR until 1929 when it became the Tajik SSR; and the Kazakh SSR and the Kirghiz SSR were created in 1936.
-
[2]
This is not to forget that the Stalinization of Central Asia also had disastrous effects. One of the most illustrative examples is perhaps the Kazakh famine of 1930-1933 which, according to a series of estimates, killed between 25% and 42% of the population of Kazakhstan (see among others Pianciola, 2001).
-
[3]
The examination of the fertility transition in the Central Asian republics by Blum (1987) starts from 1950 only. As we shall see, important changes took place before that date.
-
[4]
Given the constraints on the available data (information based on the vital registration system and from census distributions of the female population by age and marital status), a set of four indices were developed to compare the fertility experience of various European countries. For further details, see: http://opr.princeton.edu/archive/pefp/indices.aspx, last accessed 25 May 2016).
-
[5]
As the Central Asian republics did not exist at the time of the 1897 census, Central Asian republics are distinguished only from the 1926 census on.
-
[6]
The criteria used to define an urban area evolved across censuses. From an administrative criterion in 1897, the Soviet censuses moved to a size-function criterion. Despite the use of a similar basic conceptual approach to define an urban area, the minimum size varied across censuses and also across republics of the Soviet Union (for more details, see Rowland, 1986).
-
[7]
For example, in Kazakhstan, starting with the Third Five-Year Plan (1938-1941), the number of clinics grew by 144% between 1938 and 1941 and the number of delivery beds increased by 680%, reaching 14,782 in 1941. Developments were more substantial in Kazakh regions than in Russian ones, indicating that Kazakh needs were greater (Michaels, 2003).
-
[8]
Primary infertility – the proportion of childless women – is distinguished from secondary infertility – the proportion of women who have already borne at least one child but are unable to have any more.
-
[9]
Implied average parities were first computed by summing the product of PPR value and the parity reached. The contribution of the change in each PPR was then assessed by applying the PPR value at a given parity of the women born in 1915-1919 to the women born in 1935-39, thus covering the period during which fertility increased.
-
[10]
According to the Demographic and Health Surveys, 95% of the children born in the three years preceding the survey were breastfed in Kazakhstan (survey conducted in 1999), Kyrgyzstan (in 1997), Turkmenistan (in 2000) and Uzbekistan (in 1996).
-
[11]
Nomadic pastoralist communities generally have lower fertility due to higher rates of childlessness and longer birth intervals, possibly due to higher rates of early miscarriage (Henin, 1969; Roth, 1985; Ryavec, 1999).
-
[12]
“Depending on the number of children they had, women received state financial assistance that ranged from a one-time payment of 400 rubles for those with three children, to a one-time payment of 3,500 rubles plus a monthly supplement of 250 rubles to those with ten children. For each additional child after the tenth, the state offered a reward of 5,000 rubles plus a monthly stipend of 300 rubles.” (Michaels, 2003).
-
[13]
For Kazakhstan, Kyrgyzstan, and Turkmenistan, the number of births is available for four only years (1958, 1965, 1967 and 1969) between 1950 and 1970; for Uzbekistan, the number of births is available for 7 years; and for Tajikistan 10 years; other years were interpolated (Blum, 1987).
