1Before the fall of the Berlin wall, did the countries of Eastern Europe all have similar levels of fertility ? Did the official policies of gender equality and access to education result in comparable behaviours by level of education ? How did social constraints come into play ? Using census data from seven of these countries, Zuzanna Brzozowska addresses these original research questions by studying the fertility of cohorts who had their children mostly during the socialist period. She applies decomposition and standardization methods to analyse the factors affecting fertility trends, including the central role of rising educational levels. The countries in her study saw a decrease in cohort fertility and a gradual narrowing of fertility differences by level of education, although some, such as Romania, followed a more atypical pattern.
2Trends in cohort fertility in Central and Eastern Europe (CEE) [1] have not been widely discussed. Existing contributions focus mostly on contrasting the stability of state socialism with the dynamic developments during the economic and political transition that began in 1989 (Kotowska et al., 2008 ; Muresan et al., 2008 ; Perelli-Harris, 2008 ; Potančoková et al., 2008 ; Sobotka et al., 2008 ; Spéder and Kamarás, 2008 ; Stropnik and Šircelj, 2008 ; Zakharov, 2008). They usually analyse trends in period fertility and are mostly one-country studies that do not aim to provide a broader comparative perspective (Wood et al. (2014) being an exception).
3As state socialism ostensibly strove for equality in every aspect of life, one would expect rather small differences in fertility, similar to those observed in Belgium, Norway or Sweden (Andersson et al., 2009 ; Kravdal and Rindfuss, 2008 ; Neels and De Wachter, 2010). In practice, however, daily life was marked by substantial social inequalities, implying considerable educational differences in fertility, on a par with those seen in Austria or Great Britain (Prskawetz et al., 2008 ; Sigle-Rushton, 2008). The existing evidence suggests that the education-fertility relationship in CEE before 1989 was negative, but its strength varied greatly over time and across countries, fading out in Slovenia, and persisting or growing in Poland, Slovakia or Romania (Brzozowska, 2014 ; Mureş an and Hoem, 2010 ; Potančoková et al., 2008 ; Stropnik and Šircelj, 2008).
4This article examines the macro-level relationship between women’s educational expansion and completed fertility under state socialism in Europe. The aim is to determine whether the observed fertility trends were driven by changes in the educational structure or in fertility behaviour net of education. More specifically, the following questions are addressed :
- How strong was the educational gradient in completed fertility (including childlessness and high parity births) and how did it vary over time and by country ?
- What was the effect of female educational expansion on the trends in completed fertility in the analysed countries ?
- How did fertility net of growing educational attainment change ? What was the role of childlessness and high parity births in fertility changes net of education ?
- To what extent did the analysed countries share the overall and education-specific trends in completed fertility ?
5To answer these questions, we study trends in completed fertility by level of education of women born between 1916 and 1960. By choosing this timespan we cover the whole state-socialist era, stretching from the 1940s to 1989-1990. Census data from seven CEE countries are used : Croatia, the Czech Republic, Hungary, Poland, Romania, Slovakia and Slovenia. [2] These countries reflect the diversity of the CEE region, ranging from the highly secularized and relatively rich Czech Republic, through strongly Catholic and rather conservative Poland and Slovakia, to Christian Orthodox and economically less developed Romania. Croatia and Slovenia represent the more liberal Communist regime of former Yugoslavia. East Germany and the Baltic states, and the most populous country, Russia, are excluded due to a lack of data ; this is a clear limitation of the study.
6Section I of this article describes how female educational attainment and labour force participation expanded and how state socialism defined the role of women, their rights and duties as mothers, wives and employees. Section II gives an overview of country-specific social and reproductive policies. The analysis then focuses on trends in female educational structure and fertility, including childlessness and high parity births. Changes in the completed fertility rate (CFR) are decomposed into those driven by growing educational attainment (structural effects) and those due to changing fertility within the educational groups (direct effects). In Section III, the direct effects are further decomposed into parity-specific components. Several hypothetical fertility scenarios are developed in Section IV, and we assess the role of increasing educational enrolment and education-specific trends in fertility (together with childlessness and high parity births) in CFR changes. The concluding section discusses the results.
I – The situation of women under state socialism
Education and employment
7In the late 1940s and 1950s, Central and Eastern Europe underwent rapid industrialization. The economy faced severe labour shortages and desperately needed new workers (Berend, 2005). To meet growing demand, the governments vastly expanded vocational schooling, and vigorously encouraged women to take jobs as agricultural or factory workers, allowing them to occupy positions previously reserved for men – in the mining industry, for example (Fidelis, 2010). With the exceptions of East Germany (German Democratic Republic, GDR), the Czech Republic and, to a certain extent, Slovenia, state socialism was introduced into mostly rural societies where an overwhelming majority of the population had only primary education or no education at all. In the course of rapid industrialization and urbanization, millions of young people in CEE, both men and women, became literate, [3] received vocational training and moved to towns to work in both heavy and light industry (Berend, 1996). The economic activity rates of women in the 1970s were similar to those seen in Scandinavia (slightly below 70%), but far higher than in Western Europe (around 40-50%) (van Dijk and van der Lippe, 2001). Women had to enter the labour market not only for ideological reasons, but also out of necessity : wages were far too low for a family to live on a single income (i.e. that of the husband) (Stloukal, 1999). Except in Yugoslavia and, to a large extent, Poland, [4] even women who stayed in the countryside and worked in agriculture were required to take paid jobs in the system of collective farming where farmers became agricultural wage employees.
8In theory, state socialism liberated women and made them equal to men. In reality, however, policy measures were aimed at maintaining male dominance in the society. The attitude towards women’s paid work (as well as family and reproductive policy) was totally subordinated to political and economic needs (Zembrzuska, 2000). In times of rapid economic growth and labour force shortages, women were urged to take up jobs, including those typically reserved for men (usually much better paid than those generally occupied by women (Makkai, 1994)). When the economy was slowing down or the number of trained male workers was sufficient, women were banned from some male jobs (like mining) and encouraged to return to their traditional role as mothers (Haney, 2002 ; Fidelis, 2010). At home, they were expected to perform their customary role, i.e. to do all the housework. Thus, they were forced to bear a double burden (Makkai, 1994 ; Matysiak and Szalma, 2014) which was much heavier than in more affluent western countries as many of the modern conveniences of every-day life (household appliances, fully supplied and easily accessible shops, well organized public transport or wide car ownership) were lacking and part-time jobs were scarce (Stloukal, 1999). On the other hand, as Penn and Massino (2009) have pointed out, socialism offered “opportunities for educational and occupational advancement” and “new roles and identities to individuals”, especially for those coming from the lower social strata. Hence, the role of state socialism in liberating women and making it easier for them to combine work and family was ambiguous.
Combining work and family
9The image of a perfect socialist woman cultivated by state propaganda was that of a hard worker, devoted to her duties as a mother, wife and employee (Fidelis, 2010). During the first two post-war decades, maternity leave was very short as it was assumed that the state, – i.e. state-run nurseries and kindergartens – would take care of the children. Turning the slogans and promises into reality did not go smoothly, however, and day-care quality and supply were far below needs and expectations. Working women found it very hard to combine the role of mother and employee (Stloukal, 1999), especially since many had left their parental homes in the countryside to work in urban factories. As a result, period fertility declined steeply in the 1950s and 1960s (Figure 1).
Total fertility rate, by year and country, 1950-2012

Total fertility rate, by year and country, 1950-2012
10Initially, this downward trend was politically supported in countries confronted by economic difficulties and a high total fertility rate (TFR), such as Poland, where the government encouraged couples to have smaller families. In the late 1960s and early 1970s, however, when the TFR fell close to the replacement level or even below, the CEE states started to introduce pronatalist measures that were both coercive (restrictions in access to abortion) and supportive, offering more generous maternity leave, higher family allowances and privileges such as shorter waiting times for an apartment, for example (Haney, 2002 ; Stloukal, 1999). Throughout the 1970s and 1980s, the supportive instruments were generally reinforced. Childcare allowance often depended on the number of children ever born, thus encouraging women to have more than one child (Kocourková, 2002). Maternity leave schemes varied considerably by country, but generally they enabled mothers to stay at home for between one (former Yugoslavia) and three years (Hungary and Poland). Wage loss compensation usually consisted of two components : an amount proportional to the woman’s last salary that was paid for the first months after giving birth, and a flat-rate allowance that was paid for the subsequent months. For the period of maternity leave a woman’s position in her workplace was secure and the employer was obliged to take her back in her previous position. However, research on Hungary has shown that highly educated and qualified women used their full amount of maternity leave much less often than lower educated women because they found it difficult to stay away from work for three years. The reasons they gave were both career-related and financial (Haney, 2002).
Reproductive policies
11In nearly all countries of the region, abortion was legalized in 1956-57. For most of the time, trends in the number of abortions and births correlated closely and negatively (Baban, 1999 ; David, 1999 ; Dorbritz and Fleischhacker, 1999 ; Kapor-Stanulovic and David, 1999 ; Wynnyczuk and Uzel, 1999), reflecting the fact that abortion was socially accepted and widespread (Stloukal, 1999) ; [5] it was perceived by women as a primary means of birth control (Legge and Alford, 1986). Thus, changes in abortion law usually resulted in temporary TFR increases (when it was stricter) or decreases (when it became more relaxed). Cohort fertility may also have been affected.
12Modern contraceptives tended to be scarce, of low quality or legally restricted (Stloukal, 1999 ; Wood et al., 2014). Only in East Germany and Hungary were the pill and IUDs used by a significant share of women (Zielinska, 1987). Most of the state socialist regimes had a very conservative, puritanical attitude to sexual matters, very close to that of the Catholic and Orthodox Churches (Stloukal, 1999). The pronatalist policies of the 1960s and 1970s were often accompanied by courses on parenthood and family life in schools. However, these courses served mainly to convey biological knowledge of human reproduction and the socialist view of the place and importance of family and children in society ; topics such as contraception were rarely covered (Stloukal, 1999). Consequently, in most of the region ordinary people knew very little about effective contraception and relied mostly on coitus interruptus and the rhythm method (Berent, 1970 ; Łobodzińska, 1995).
Implications for the education-fertility relationship
13How could such a socioeconomic setting have affected the educational gradient in completed fertility ? Theoretical considerations on the education-fertility relationship have been thoroughly discussed in demographic research (Wood et al., 2014), so only the arguments directly related to this study will be mentioned here.
14As stated above, the evidence suggests that in the state socialist CEE countries the relationship between fertility and education was negative. It is often argued that because the roles of parent and student (and then employee starting her career) are perceived as incompatible, better educated women start family formation late and have fewer children (Blossfeld and Huinink, 1991 ; Kravdal and Rindfuss, 2008 ; Ní Bhrolcháin and Beaujouan, 2012). However, under state socialism even university degree holders became mothers mostly in their mid-twenties (Kantorova, 2004 ; Kreyenfeld, 2004), and could therefore have achieved larger family sizes.
15For women active in the labour market, the opportunity costs of having children are usually higher for the better educated (Becker, 1991 ; Blossfeld, 1995) as they tend to have more satisfying and better paid jobs. In state socialist countries we would expect rather small opportunity costs because of absence of competition in the labour market, almost full employment with very stable jobs, and rather limited career possibilities for female university graduates (Sobotka, 2003). All these factors, together with the relatively widespread availability of institutional childcare from the 1970s (Wood et al., 2014), made it quite easy to combine work with family life.
16Moreover, as returns to education were low (Sobotka, 2011 ; van Dijk and Lippe, 2001 ; Wood et al., 2014), there was little incentive for parents to invest in their children’s education. Thus, taking economic motivation into account, the quality-quantity trade-off, which can explain the negative gradient in fertility in meritocratic countries (Becker, 1960 ; Hanushek, 1992), is unlikely to have operated strongly under state socialism. However, it could have emerged for ideational reasons : unlike income, lifestyle differed considerably across social strata (Domański, 2004 ; Szelenyi, 1978). So it is possible that better educated women preferred to have fewer children (Mureş an and Hoem, 2010 ; Sobotka, 2004) or that they controlled their fertility more effectively (Kravdal and Rindfuss, 2008 ; Sobotka, 2004).
II – A diverse range of social policies
17The policy goals set by the states behind the Iron Curtain were very similar due to the shared ideological roots of their respective communist parties, but the means varied substantially. The governments were afraid of labour shortage ; they also believed in a direct link between national power and population size, and so procreation was proclaimed as the duty of every loyal citizen (Dorbritz and Fleischhacker, 1999). People were expected to marry early and have children quickly. The control of demographic phenomena was generally considered vital to the success of development strategies in planned economies (Kligman, 1998). Thus, population policy was highly politicized and prone to nation-specific ideological adjustments. Consequently, geopolitical, economic and cultural conditions resulted in a multifaceted socialist reality that varied considerably over time and across countries.
The Czech Republic and Slovakia
18Czechoslovakia tried to stimulate fertility through positive incentives. The first modest pronatalist measures were introduced in the late 1950s, and by the early 1960s women were entitled to 22 weeks of maternity leave at 90% of salary. In the late 1960s and early 1970s the government introduced further pronatalist incentives, with increases in maternity cash payments and family allowances, an extension of maternity leave to 26 weeks, paid post-maternity leave up to the child’s second birthday for women with at least two children (Kocourková, 2002), followed by low-interest housing loans for young married couples (McIntyre, 1975). Obtaining an abortion was quite easy, [6] although it required the consent of local abortion committees until 1987, when abortion became accessible upon request. In Slovakia, women terminated pregnancies less often than in the Czech Republic, but in both regions abortion remained an important instrument of fertility control (Wynnyczuk and Uzel, 1999).
19By the 1970s, married couples with children in Czechoslovakia received some of the most generous social benefits available behind the Iron Curtain. Having children almost guaranteed access to a rental apartment, entitled parents to substantial allowances and payments, and made women eligible for earlier retirement (differentiated by the number of children ever born). All in all, including all benefits and subsidies for nurseries and kindergartens, school meals, transportation, children’s clothing, etc., Czechoslovakia spent 10% of its 1975 budget on population policy measures (Frejka, 1980). This probably helps to explain why the country had one of the region’s lowest proportion of childless women and one of the highest percentages of ever married people (Wynnyczuk and Uzel, 1999). Thanks to the generous benefits and subsidies, the TFR stopped falling and rose to 2.4 children per woman in the early 1970s before declining again (Figure 1). Cohort fertility, on the other hand, decreased monotonously from around 2.2 in the 1920 cohort to 2.0 among women born in 1960. However, it might have fallen even lower without the supportive measures introduced (Frejka, 1980).
Hungary and Poland
20In 1959, in response to sub-replacement fertility rates, the Hungarian government introduced allowances aimed at increasing the progression to second and third births. During the 1960s and 1970s, financial benefits were increased. Maternity leave lasted for five months, was paid at full salary (McIntyre, 1975) and closely resembled the Polish scheme established in 1972. In both countries, post-maternity leave was introduced in the late 1960s and expanded in the early 1970s (to 31 flat-rate paid months in Hungary and 36 unpaid months in Poland) (Frątczak et al., 2007 ; McIntyre, 1975). They were aimed not only at raising fertility rates but also at removing women from the labour market (although the female labour force participation was rather modest compared to other CEE countries) (Haney, 2002 ; Kurzynowski, 2000 ; Zembrzuska, 2000 ; Matysiak and Szalma, 2015). The planned economies slowed down in the 1970s and were unable to continue providing full employment. As unemployment did not officially exist under state socialism, governments focused their efforts on encouraging people to leave the labour market. In Poland, these efforts peaked during the economic agony of the early 1980s when paid post-maternity leave was extended to 24 months [7] (Frątczak et al., 2007 ; Titkow, 1999). As a result, period TFR rose slightly from 2.3 in 1980 to 2.4 in 1983, but then started to decline again, falling to around 2.0 in 1989 (Figure 1 ; Kotowska et al. 2008).
21Legalized in 1956, abortion was easy to obtain in both countries on medical and social grounds. [8] In 1974, Hungary introduced some restrictions for married women with fewer than two children living with their husbands in adequate housing (McIntyre, 1975), but the law was not strictly enforced (David, 1999). In Poland the abortion rates were rather low by CEE standards, but in Hungary they remained high (David, 1999 ; Titkow, 1999), despite campaigns fostering effective contraception, and easy, low-cost access to modern contraceptives. According to surveys conducted in the 1970s, nine in ten sexually active women under age 35 practiced some form of contraception. Almost half of them were taking the pill and nearly 20% used an IUD or condoms (David, 1999). Such high take-up of effective contraception was not seen in any other CEE country except East Germany.
Romania
22In postwar Romania, period fertility started to fall in the mid-1950s (Figure 1). In 1964, seven years after introducing the most liberal abortion law in Europe (abortion was available upon request), around 80% of pregnancies in Romania were terminated artificially (Baban, 1999). In view of these skyrocketing abortion numbers and falling fertility (TFR dropped from 3.2 in 1950 to 1.9 in 1966), in late 1966 the government implemented a range of draconian pronatalist measures : abortion became severely restricted, [9] divorce was allowed only in exceptional cases (and was very costly) and childless persons over age 25 had to pay a special tax of 10-20%, depending on their income (Moskoff, 1980). These coercive measures were accompanied by incentives such as higher family allowances, expanded childcare facilities and opportunities for part-time employment for mothers (McIntyre, 1975), which had little practical impact. The cash benefits amounted to less than one-tenth of an average salary, nurseries were of very poor quality and very few women could afford to reduce their working hours (Moskoff, 1980). Moreover, maternity leave was one of the shortest of any CEE country (16 weeks), female labour force participation rates were among the highest in Europe (almost 80% in 1956) and the dual burden placed on working women was particularly high as modern household appliances were lacking and everyday life involved queuing even for basic goods (Baban, 1999 ; Moskoff, 1980).
23The restrictive 1966 policy together with the unavailability of modern contraceptives [10] led to an upsurge in the TFR, which almost doubled to reach 3.7 in 1967. Afterwards, however, when enforcement of the anti-abortion law became less strict, fertility dropped again, reaching around 2.6 in 1975, and continued to decline to about 2.2 in 1989 (Figure 1). Thus, in the long run, fertility levels in Romania did not exceed those seen in other CEE countries, reflecting the long-tern failure of the government’s extreme pronatalist measures.
Croatia and Slovenia
24In contrast to other state-socialist countries, Yugoslavia had a labour surplus, so the authorities were not strongly motivated to secure full employment (Antic, 1992), and fewer efforts were made to attract women into the labour market : female labour force participation rates were around 40% in 1981 (reaching 60% in Slovenia) (Reeves, 1995). Although the TFR was declining (and has remained below 2 in Croatia since 1970 and in Slovenia since 1980), Yugoslavia did not implement such vigorous pronatalist incentives as other CEE countries. Parental leave lasted between 15 and 30 weeks and could be extended until the child was 8 or 12 months old when combined with part-time employment (Kapor-Stanulovic and David, 1999). In addition, extensive maternity benefits were introduced in 1978 (Antic, 1992).
25From 1951 abortion was authorized on socio-medical grounds, and in 1960 broadly defined social reasons were included. Modern contraceptives started to become more readily available in 1969, but a 1970 survey found that only half of married women practiced contraception, of whom almost 70% relied on withdrawal (David and Kapor-Stanulovic, 1999). Abortion was widely practiced until the 1980s, but less so from then on.
III – Data and Methods
Data
26We used data from population censuses in the following countries : the Czech Republic (1980, 2001), Croatia (2001), Hungary (1980, 1990, 2001), Poland (Fertility Survey accompanying the 2002 census), Romania (1992, 2002), Slovakia (2001) and Slovenia (2002). All data were taken from the Cohort Fertility and Education Database [11] (Zeman et al., 2014) ; those for Hungary, Romania and Slovenia were derived from IPUMS International [12] and were 5% (Hungary) and 10% (Romania and Slovenia) census samples. [13] For data drawn from more than one census an arithmetic mean was computed. [14] Completed fertility was derived from the total number of live births per woman. Women aged 40-76 at the time of the census were included in the analysis. [15] This corresponds to the 1916-1960 birth cohorts, which were analysed in fiveyear groups from 1916-20 to 1956-60. [16] We thus study the (almost) completed fertility of women whose reproductive careers mostly took place under state socialism. [17]
27Four educational categories were used for the analyses, corresponding to the following levels of the International Standard Classification of Education 1997 (ISCED-1997) : [18] primary and lower (ISCED 0, 1 and 2), basic vocational (ISCED 3C), secondary (ISCED 3AB and 4) and tertiary (ISCED 5 and 6). In Hungary, the basic vocational category emerges only among women born after 1940, but for the sake of comparability with other countries we kept the four-category classification. The educational systems were similar across the region. Primary school usually lasted for eight to nine years and was followed by differentiated upper secondary education with three main tracks : lower vocational (labelled here as basic vocational), secondary general and upper vocational, merged here into one category, secondary education (Kogan, 2008). Unlike in the Soviet Union, the post-secondary sector in the analysed countries consisted mainly of university education and was rather exclusive (Kogan, Gebel and Noelke, 2012). Under state socialism the difference between basic vocational and secondary education was important, corresponding to that between manual and clerical workers. The following section clearly shows that these two groups differed in their demographic behaviour.
28Cases of unknown education were excluded from the analyses. The exact sample sizes and missing cases are given in Appendix Table A.1.
Methods
29First, we employed decomposition analysis drawing on Kitagawa’s (1955) and Cho and Retherford’s (Canudas Romo, 2003) work. Fertility changes over consecutive cohorts were decomposed into structural (1a) and direct (1b) effects.
31which denote the proportion of educational group i in the cohorts t and t + h, respectively.
32In order to assess the effect of parity changes (including childlessness), the direct effect (1b) can be further decomposed into two components measuring the effect of standardized mother’s fertility by parity (2a) and standardized childlessness (2b) :
34where : MCFR(t)ji and MCFR(t + h)ji denote mothers’ completed fertility rate of parity j in educational group i at time t and t + h, respectively, and p(t)i and p(t + h)i denote the share of mothers in educational group i at times t and t + h, respectively. Thus, 1 – p(t)i and 1 – p(t + h)i are the proportions of childless women.
35In addition, direct and indirect standardization were employed to develop different fertility scenarios. The fixed-education scenario is a directly standardized CFR computed as follows :.
37The classic indirect standardization (i.e. the fixed-fertility by education) is calculated as :
39Noticing that
41formula (4) can be easily transformed into a scenario of fixed childlessness :
43Finally, to assess the importance of changes in high-parity births we computed conditional CFR3+ which tells us how many offspring an average mother of at least three children had, and is defined as :
45where C denotes the number of all children, C1 andC 2 stand for the number of children of birth order one and two, respectively ; F is the total number of women, F0, F1, F2 denote the number of women with zero, one and two children, respectively. CFR can be then defined as :
47where MCFR1 and MCFR2 denote mothers’ CFR of parity one and two, respectively, and is the proportion of women with three or more children.
48Thus, the respective standardized fertility scenario (with CondCFR3+ and being kept constant) was computed in the following way :
50In all the scenarios, the first cohorts, for which data are available, were used as a reference. Thus, t is constant. All computations were performed in the statistical package R.
IV – Results
Educational expansion and completed fertility
51Figure 2 shows the continuous expansion in female education that occurred in all countries. In the youngest cohorts, the proportion of women with only primary education is two to three times lower than in the oldest cohorts. Among women born between 1916 and 1930 in Croatia, Hungary or Romania, at least 80% had primary education only. In contrast, among women born in the 1950s the share was below 40% ; in the Czech Republic, Poland and Slovakia it was below 20%. By the 1946-50 cohort, more than half of the female population had at least basic vocational education (except for Romania where this proportion was achieved in the next five-year cohort). In most countries it was secondary education that expanded most, becoming the modal level of education in the youngest cohorts, while in the Czech Republic and Romania, basic vocational education was the predominant level of attainment. With time, university education became less exclusive, although it still concerned less than one-fifth of the female population in the youngest cohorts.
Educational structure (left-hand axis), completed fertility rate and mothers’ completed fertility rate (right-hand axis) by birth cohort and country

Educational structure (left-hand axis), completed fertility rate and mothers’ completed fertility rate (right-hand axis) by birth cohort and country
52There was a general downward trend in completed fertility, although initial levels varied greatly across countries (Figure 2). While in the Czech Republic, Hungary and probably Slovenia, women in the 1916-20 cohort had a completed fertility of around 2.2 children per woman, among those born ten years later in Poland and Slovakia it was still considerably higher, at 2.7 and 2.8, respectively. In the youngest cohorts, the differences narrowed, with completed fertility ranging from 1.9 to 2.2 (except for Slovenia with 1.75). In Slovakia and Slovenia the average number of children per woman decreased monotonously (though at very different speeds) ; in Hungary and Poland it started to rise in the more recent cohorts, followed by the youngest cohort in Croatia. The Czech Republic experienced only a minor fertility reduction. Romania showed its own unique pattern : fertility fell among women born between 1916 and 1930, went up again until the 1941-1945 cohort, and then resumed its decline.
53In all countries but Slovakia, the difference between CFR and mothers’ CFR (MCFR) decreased with time (most noticeably in Romania). This reflects the falling childlessness rates, which converged across the countries in the youngest cohorts.
Trends in completed fertility by education
54All countries exhibited a negative educational gradient in completed fertility and usually a positive one in childlessness. Figure 3 shows all three possible trends in educational differences in fertility : convergence, stability and divergence. Convergence took place in the Czech Republic, Slovakia and Slovenia, where it was most clearly visible (the difference in CFR between the least and the highest educated dropped in the youngest cohort to 0.3 in absolute values or 20% in relative terms). In contrast, in Slovakia the curves narrowed only slightly ; they retained a strong education gradient, varying by 1.1 children between those with primary and tertiary education (or over 60% in relative terms).
Completed fertility rate by education, country and birth cohort

Completed fertility rate by education, country and birth cohort
55In both Poland and Hungary the educational differences in fertility remained stable over time, but they varied strongly in extent : while they were moderate in Hungary (and similar in magnitude to those observed in the Czech Republic and Croatia in the last cohorts), they remained very large in Poland, reaching almost 1.3 children between the extremes (with low educated women having 80% higher fertility than graduates). Probably, the rise in fertility seen in Hungary among the least educated born after 1940 was driven by the emergence of the additional category of basic vocational education. As a result, the group with primary education became more selective.
56Finally, in Croatia and Romania, the gap between the lowest educated and all the others became smaller, but the differences between those with basic vocational, secondary and tertiary education increased from the 1946-50 cohort. Throughout the observed cohorts, the educational differences in fertility in Romania were the widest of all countries. Among the youngest women, the least educated had over 1.5 (or 120%) more children, on average, than university graduates (Figure 3).
57In contrast to completed fertility, trends in childlessness were very similar in all countries and in all educational groups (Figure 4). The proportion of ultimately childless women declined with time. While among women born in 1916-1920 levels of childlessness varied from 12% (the Czech Republic) to 23% (Romania), among those born 40 years later the values ranged from 4.5% to 10%. The educational differences did not vanish, but they grew smaller with time. For the youngest cohorts in most countries a clear distinction is visible only between the highly educated and the rest, with an average difference between these two groups of six percentage points (varying from three in the Czech Republic to eight in Croatia and Romania).
Proportion of ultimately childless women by education, country and cohort

Proportion of ultimately childless women by education, country and cohort
58Childlessness fell most quickly in the prewar cohorts ; in the younger ones the downward trend slowed down (the Czech Republic and Hungary), levelled off (Poland and Romania) or reversed (Croatia and Slovenia). In most countries, the turning point occurred in the 1941-45 cohort. Only in Slovakia, where the proportion of the ultimately childless was already very low in the oldest cohorts, did it remain relatively stable.
59In the analysed cohorts, the overall share of mothers of three and more children shrank from 40-50% to around 30% in Poland, Romania and Slovakia, 20% in Croatia, the Czech Republic and Hungary, and below 15% in Slovenia (Figure 5). However, the decline was neither continuous nor universal across the educational groups. In the Czech Republic, Hungary, Poland and Romania the proportions tended to fall in the older cohorts and then to rise or stabilize in the younger ones. In Slovenia they halved among the least educated, approaching the values seen in other groups, while Slovakia experienced minor reductions across all educational strata. In Croatia, the proportion increased among all except the least educated. Apart from young cohorts in Slovenia, the proportion of mothers of three and more children was strongly differentiated by education, with the highest values among the least educated and the lowest among university graduates. Between women with tertiary, secondary and basic vocational education, the differences were generally moderate ; they were largest with respect to primary education. In Poland and Slovakia, however, the differences were equally large between all educational groups.
Proportion of mothers of three or more children by education, country and cohort

Proportion of mothers of three or more children by education, country and cohort
60All in all, the decrease in the proportion of mothers with three and more children was largely driven by changing educational composition. A similar conclusion can be drawn for the decline in the average number of children among women with at least three offspring (Figure 6). It fell typically from over 4 in the oldest cohorts to around 3.5 in the youngest ones. The highest figures were observed for Romania, starting with almost 6 and sinking to slightly above 4. In Slovenia, the conditional CFR3+ not only decreased (from 3.6 among women born in 1926-30 to 3.1 among those born 30 years later), but also converged across the educational groups. Thus, having more than three children became very rare in Slovenia in all educational strata. In all countries, the negative educational gradient became weaker with time. It remained strong in Poland, Romania and Slovakia, however, with an average difference among the youngest women of over one child between the lowest and highest educated.
Average number of children per woman with at least three children (conditional CFR3+), by education, cohort and country

Average number of children per woman with at least three children (conditional CFR3+), by education, cohort and country
Changes in fertility : structural and direct effects
61Figure 7 shows how growing educational attainment (structural effect) and changes in fertility within the educational groups (direct effect) affected cohort fertility. In all countries and cohorts, increasing education lowered fertility. This dampening effect varied greatly in strength, but in most countries it peaked for women born between 1936 and 1950, i.e. for those who massively entered secondary education between the early 1950s and the mid-1960s. The direct effects usually correspond with the overall changes in CFR. Only in Slovakia were direct effects persistently negative and usually outweighed by the effect of education. In other countries, the two effects varied in strength from cohort to cohort, with neither influence clearly dominating in the long run.
Completed fertility rate (right-hand axis) and direct and structural effects of changes in completed fertility rate (left-hand axis), by cohort and country

Completed fertility rate (right-hand axis) and direct and structural effects of changes in completed fertility rate (left-hand axis), by cohort and country
62When the direct effect is decomposed, it is found to comprise several counteracting forces (Figure 8). While declining childlessness and the growing proportion of women with two children (Appendix Figure A.1) consistently raised the CFR (net of structural changes), births of parity three and more among women born before the 1950s mostly moved it in the opposite direction. In the younger cohorts exposed to the pronatalist measures of the 1970s and 1980s, fertility at high parities tended to increase, exerting a positive effect on completed fertility within the educational groups. This trend is most striking in Romania, where high parity births increased strongly among women most affected by Ceauşescu’s coercive policies, regardless of the changing educational structure. In all countries, the proportion of mothers of only one child tended to decline within the educational groups, especially in the 1940s cohorts.
Direct effect decomposed into the effects of childlessness and mothers’ parity-specific completed fertility, by cohort and country

Direct effect decomposed into the effects of childlessness and mothers’ parity-specific completed fertility, by cohort and country
Scenarios of completed fertility
63Employing direct and indirect standardization gives various hypothetical fertility scenarios that answer the question of how completed fertility would have evolved if some of its components had not changed. As expected, fertility would have been higher in all countries if educational attainment had not increased (Figure 9) : in Poland and Romania, an average woman born in the late 1950s would have had 0.5-0.6 more children. Similarly, in Slovakia, Hungary and the Czech Republic, the fertility of the youngest cohort would have been higher than that of the oldest cohorts, and in Croatia it would have been at the same level. In Slovenia, however, growing educational attainment barely influenced completed fertility, which was lowered much more substantially by reductions in high parity births (net of education). Had high parity births not fallen, fertility would have been higher in the youngest cohorts by over 0.2 children, thus reaching almost 2 ; in Slovakia the difference was only slightly greater (0.25). In other countries, the standardized curve had an inverted U-shape ; it went up and then down, moving closer to the observed CFR. Such a development demonstrates that, net of education, high parity births dropped sharply among the older cohorts but increased among the younger ones. The strongest decreases were seen in Romania and Hungary, where an average woman born in the early 1930s would have had 0.7 more children if conditional CFR3+ and the proportion of women with three or more children had stayed the same.
Fertility scenarios by cohort and country

Fertility scenarios by cohort and country
64Reductions in childlessness exerted a positive effect on fertility in all countries. If the levels of childlessness had not changed, fertility in the youngest cohorts would have been approximately 0.15 children lower in Croatia, Poland and Slovenia, 0.2 lower in the Czech Republic, 0.3 in Hungary and 0.5 in Romania. As expected, it would have had no effect in Slovakia, where the proportion of childless women changed very little over cohorts.
65Finally, classic indirect standardization, i.e. keeping the education-specific fertility rates fixed at the level of the oldest cohorts, shows that fertility net of educational expansion increased by up to 0.4 children between the oldest and the youngest cohorts. The only exception was Slovakia, where it barely changed. In Croatia, Hungary, Poland and Romania, fertility declined among the older cohorts, but for women born in the late 1940s and in the 1950s it clearly went up net of education, exceeding the levels seen in the oldest cohorts.
Conclusion
66Despite the usual practice of treating the state socialist countries en bloc, and despite their similar political and economic systems, there is only limited evidence for the existence of a single “socialist” fertility pattern. The general trends appear to have corresponded to those in the West, namely a decline in fertility driven largely by growing educational attainment among women and falling high parity births. Two became the modal number of children in the family (for CEE, see Appendix Figure A.1) and the two-child ideal was shared across Europe (Sobotka and Beaujouan, 2014). Thus, the quantitative changes in fertility in the second half of the twentieth century should be seen as common to the whole of industrialized Europe rather than as specific to market democracy or state socialism (Monnier and Rychtaříková, 1992). When inspecting the trends more closely, a socialist peculiarity does emerge, that of very low levels of childlessness (and, not considered here, the low age at first birth ; e.g. Sobotka, 2011).
67While in Western Europe the prevalence of childlessness declined only among women born before the 1940s, and was usually followed by an increase (Beaujouan et al., 2015), behind the Iron Curtain it continued to fall across all the analysed cohorts, or stabilized at a very low level. This trend was likely closely linked to social policies adopted under state socialism, which strongly encouraged citizens to marry early and become parents quickly. The policies reflected and supported the social expectations associated with reaching adulthood : finishing education, finding a job, getting married and having children (Sobotka, 2004). Following such a trajectory was made easier by the low opportunity costs of motherhood. There were few career opportunities incompatible with family life, especially for women, who rarely had to choose between job promotion and having children.
68Furthermore, the importance of family under state socialism should not be forgotten. To a certain extent it corresponded to the golden age of marriage in Europe and the US in the 1950s and 1960s (Cherlin, 2004 ; Festy, 1980 ; Kuijsten, 1996). But family fulfilled an additional function in a non-democratic system with restricted freedom and extensive control over behaviour at work and in public places : it provided a sphere of individual freedom, a sort of refuge (Andorka, 1997 ; Sobotka, 2003).
69In all the analysed countries, fertility within the educational groups tended to rise, mostly in the younger cohorts exposed to pronatalist incentives. At population level, however, the increases were barely visible (especially in the older cohorts) as they were cancelled out by growing educational attainment, which consistently lowered observed fertility. Thus, our results do not disprove the effectiveness of the policy measures adopted under state socialism.
70It might seem surprising that in countries led by communist parties and driven by egalitarian ideology, a strong negative educational gradient in fertility persisted. The communists who took power called not only for social and gender equality, but also for empowerment of workers and farmers. Yet, the educational differences seen in most countries (except for Slovenia in the youngest cohorts) are no smaller than those in Western Europe (Andersson et al., 2009 ; Neels and De Wachter, 2010 ; Prskawetz et al., 2008 ; Sigle-Rushton 2008). Why did they not vanish in the course of time ? As explained above, opportunity costs could not have been the reason, because they were small. However, although income inequalities were very moderate among ordinary people (Bafoil, 2009), they were profound and persistent in social terms : household studies found significant variations in household amenities across social strata (Andorka, 1995), and large differences in lifestyle were observed, particularly among the highly educated (Domański, 2004 ; Giza-Poleszczuk, 2007 ; Szelenyi, 1978, 1982). Contrary to the official ideology and policy, the higher social strata behaved as predicted by Becker’s theory of raising high-quality children (Becker, 1991). Considering the low earnings and harsh state-socialist reality, this made it very difficult to have a large family and led to a quantity-quality trade-off. Besides, better educated women tended to use modern contraceptives more often (David, 1999 ; Kapor-Stanulovic and David, 1999), or, as the Romanian case shows, they either knew how to employ the natural methods of birth control more effectively, were able to get contraceptives on the black market or from abroad, or had better access to illegal abortion.
71While the fertility trends behind the Iron Curtain followed a similar pattern, their strength differed enormously. Generally, one rule seems to prevail : the higher the levels of fertility in the oldest cohorts and the steeper its decline across cohorts, the wider the educational differences in fertility and the stronger the effect of changing educational structure on CFR.
72Applying these criteria, the analysed countries can be divided into two groups. The first and bigger one consists of the richest and most developed countries, i.e. Croatia, the Czech Republic, Hungary and Slovenia. In these countries, fertility was already relatively low in the oldest cohorts (around or below 2.2) and it fell only moderately to 1.8-2.1 despite profound changes in the educational structure. The educational gradient in fertility was weaker than in the other countries and it tended to decrease with time.
73The second group consists of Poland and Slovakia, where both the decline and the initial educational gradient in fertility were very steep, while the proportion of childless women fell only slightly (Poland) or not at all (Slovakia). Between the 1926-1930 and 1956-1960 cohorts, fertility fell from around 2.8 to 2.2 children per woman, thus reaching only slightly higher levels than the first group. The fact that the Czech Republic and Slovakia belong to different groups might seem surprising, as until 1993 they formed one federal state and implemented the same policies. This is an interesting example of how the different initial population structure, culture and stage in the demographic transition [19] might differentiate responses to the same policy measures and broader institutional settings.
74The fertility pattern in Romania reflects the drastic policy measures imposed by Nicolae Ceauşescu in the mid-1960s, and defies classification. Women born between 1916 and the mid-1930s experienced a rapid decline in completed fertility, driven mostly by very strong reductions in high parity births, while the following cohorts of younger women born before the 1950s (i.e. those most affected by Ceauşescu’s pronatalist policy) tended to have more children and to remain childless less often ; later the figures stabilized. At the population level, however, the increases in fertility are visible only for the 1936-1945 cohorts ; afterwards, the rapid rise in educational attainment resulted in fertility decline. The educational gradient in completed fertility remained strictly negative and was the strongest of all analysed countries. It seems that in a country with low standards of living, very strict anti-abortion law and family planning limited to natural methods, education went hand in hand with effective fertility control.
75In sum, the developments in completed fertility of women born between 1916 and 1960 were similar on both sides of the Iron Curtain. A specific socialist fertility trend does not seem to have existed, with one exception : a steady decline in the proportion of childless women (contrasting with an upward tendency among women in the West born after 1940). The extent of fertility changes varied considerably across the CEE countries under state socialism, as did the role of structural and direct components. The results also suggest that in Romania, Poland and Hungary, strong pronatalist policy measures may have halted the decline in high parity births.
Acknowledgements
This research was funded by the European Research Council under the European Union’s Seventh Framework Programme (FP7/2007-2013) / ERC Grant agreement 284238 (EURREP project). I am grateful to Tomáš Sobotka, Éva Beaujouan, the Editors of Population and three anonymous reviewers for their valuable comments on an earlier version of this paper, and to Kryštof Zeman for providing part of the data.


Notes
-
[1]
In this article, the term CEE refers to the political entity, that of the former state socialist countries in Europe. It does not include the USSR. In geographical terms, the countries studied in this article belong to Central Europe (the Czech Republic, Hungary, Poland, Slovakia) and Southeastern Europe (Croatia, Romania and Slovenia).
-
[*]
Wittgenstein Centre for Demography and Global Human Capital, Vienna Institute of Demography, Warsaw School of Economics.
Correspondence : Vienna Institute of Demography, Austrian Academy of Sciences, Welthandelsplatz 2, Level 2, 1020 Vienna, Austria, tel. : 0043 (0) 1313367750, email : zuzanna.brzozowska@oeaw.ac.at -
[2]
To avoid confusion, the current names of the states are used. Thus, the Czech and Slovak parts of Czechoslovakia are referred to as the Czech Republic and Slovakia, respectively.
-
[3]
For example, in Poland, 23% of the population was illiterate in 1931. Thanks to the anti-illiteracy campaign organized by the government, mostly between 1949 and 1952, the proportion had fallen to 2.7% by 1960 (Łepkowski, 1964, 1973).
-
[4]
Collective farming existed in Poland, but it covered only 18.5%, at most, of the total farmland (Encyklopedia PWN, 2014). In Yugoslavia, post-war collectivization had almost completely disappeared by 1954 (Wädekin, 1982).
-
[5]
Abortion was probably not evenly spread across social strata. However, the few available data that break down abortion numbers by education give contradictory results for the educational gradient in abortion : it was negative in Hungary, but positive in Poland (David, 1999 ; Okólski, 1983).
-
[6]
Between 1962 and 1964, childless women had limited access to abortion (McIntyre, 1975), and in 1973 the government imposed restrictions for women with fewer than three children. However, the broadly defined social reasons for authorizing abortion ensured that the restrictions had a very limited effect (Wynnyczuk and Uzel, 1999).
-
[7]
In all, (post-)maternity leave lasted for three years ; the last year was unpaid, but, like the first two, counted towards the pension entitlement.
-
[8]
Since 1993, only abortions for medical or crime-related reasons have been authorized.
-
[9]
Abortion was authorized for women above age 45, those who already had at least four children, those whose life was endangered by pregnancy, those whose pregnancy was a result of crime (rape or incest) and those who were physically, psychologically or emotionally incapacitated (the medical indications were listed in detail). Later the age threshold was changed (to 40 in 1973 and back to 45 in 1985) and the required number of previously born children increased to five in 1985 (Baban 1999). Illegal abortions were widespread, but risky and very expensive (Moskoff, 1980). The abortion law was liberalized in 1990 (Baban 1999).
-
[10]
They were neither produced domestically nor imported (Moskoff, 1980).
-
[11]
Cohort Fertility and Education Database, http://www.cfe-database.org.
-
[12]
Data originally produced by the statistical offices of Hungary, Romania and Slovenia.
-
[13]
For more details on the data sources for each country, see the documentation files in the CFE database, http://cfe-database.org.
-
[14]
When using two or more data sources we first validated their quality by comparing trends in the educational structure and in overall and education-specific fertility. Only censuses that were fully consistent with each other were included in the analysis.
-
[15]
For countries with more than one available data source, cross-checks proved that the chosen age range does not cause a selectivity bias. Completed fertility of women aged 40 at the time of one census did not differ from that reported by women aged 50 ten years later in the next census. Also, women aged above 70 did not seem to be affected by selective survival : their educational structure and education-specific fertility were very similar to those traced 10 or 20 years back.
-
[16]
The Slovenian cohorts are shifted by one year, starting from 1917-1921 and finishing with 1957-1961, as the five-year age groups in the original data do not fully correspond to the cohort categories used in this study.
-
[17]
The data used do not take emigration into account. However, we do not expect this limitation to have significantly distorted the results. Emigration under state socialism occurred usually in waves (e.g. in the late 1960s in Czechoslovakia or in the 1980s in Poland) with the departures of mostly young (under age 40) and well-educated people (Maatsch, 2012). Thus, if emigration had been an important factor for the analysis, there would have been some spikes in the data. None were observed.
-
[18]
International Standard Classification of Education developed by UNESCO. For more information, see http://www.uis.unesco.org/Education/Pages/international-standard-classification-of-education.aspx.
-
[19]
When communist Czechoslovakia was established in 1948, the Czech Republic had already completed the demographic transition, while it was still in progress in Slovakia.