The mid-twentieth century fertility boom from a global perspective

Abstract

During the central decades of the twentieth century there is ample though often indirect evidence that a significant rise in fertility took place in much of the world. In some countries with historic demographic transitions this trend change has been called the baby boom. Elsewhere it has often been called the demographic explosion. Seldom has it been addressed from a global perspective. The main goal of this paper is to study these shifts comparatively, assessing the extent to which the timing and the mechanisms behind increasing fertility were or were not shared by different areas of the world. The paper provides a detailed description of fertility trends in 13 countries from four continents, based on a cohort approach to fertility and making use of data from the Integrated Public Use Microdata Series, International (IPUMS-I). Our analysis shows that, with some exceptions, increasing fertility was a global demographic phenomenon, although there are important variations in terms of intensity, timing, and duration.

Keywords::

• baby boom
• cohort analysis
• developed countries
• developing countries
• fertility
• fertility cycles
• education

Notes

^1. Using the World Fertility Surveys data as the primary source for this exercise is not feasible. The WFS, excellent as they may be, are not appropriate for reconstructing the fertility of cohorts born before 1930 or 1935, while census data allow us to track fertility for cohorts born as early as 1900.

^2. Certain aspects of these data warrant our attention. Non-response poses a potential problem though generally levels are relatively low. Only in Morocco are levels of non-response above 5% in any age group. This variable is not correlated with age except in the case of Morocco where it reaches 5% of the oldest age groups. There is no indication that the quality of response decreases significantly with age. In one specific study of this matter dealing with Latin American nations the opposite has been shown to happen (see Garcia et al, 2013). In Morocco there was a problem with women’s date of birth, missing in nearly 24% of the cases. We decided against imputing year of birth from age on the census returns because ages reported varied widely, suggesting that the lack of information regarding date of birth was ultimately a problem of numeracy among those women and that fertility by age would not be reliably reported either. In any case, CEB by birth cohort would have changed only slightly even if these new cases had been included. Data for the 1905–1914 birth cohorts in Morocco and < 1905 birth cohort in Costa Rica have not been included because of very small numbers and data instability. In Spain, an adjustment of census data to vital statistics in higher parities was necessary to overcome the problem of top-coding imposed on ‘the children ever born’ variable in the 1991 Census. The same situation was found in Morocco and Turkey where the CEB variable was also top-coded (13 in Morocco, 10 in Turkey). In both cases, actual numbers of childbirths were adjusted by means of the parity distributions holding for women 45–49 on the first round of the WFS. These adjustments were applied to all cohorts in each census. For evident reasons, the new CEB estimates in each country increased slightly.

^3. Censuses used in this work are: USA 1990 (N = 12,501,046), Argentina 2001 (N = 3,626,103), Costa Rica 2000 (N = 381,500), Mexico 2000 (N = 10,099,182), Austria 1991 (N = 780,512), East Germany 1981 (N = 4,278,563), Hungary 1990 (N = 518,240), Spain 1991 (N = 1,931,458), Greece 1991 (N = 951,875), China 1982 (N = 10,039,191), Turkey 1985 (N = 2,554,364), and Morocco 2004 (N = 1,482,720).

^4. See also Lee and Feng (1999) for similar trends depicted by fertility period indicators between 1950 and 1982. For an assessment of the 1982 Chinese Census, see Banister (1987).

^5. In Turkey, the only census in IPUMS-I to pass our quality control tests was carried out in 1985. While yielding very interesting data for early cohorts, there are little data for more recent ones.

^6. The following table contains the specific results of this test

^7. For Spain, see Requena (1997), Reher, Requena, Sanz-Gimeno, and Sanchez-Dominguez (2013) and Requena and Salazar (2014); for the USA, see Ryder (1986); for Latin American countries, see Rosero-Bixby (1996), Fussell and Palloni (2004) and Pérez Brignoli (2010) and for a sample of several developing nations (Dyson & Murphy, 1985).

^8. Period total fertility rates tended to be slightly higher than completed cohort fertility in developed and the developing nations used here, with differences ranging from 0.1 to 1.2.

^9. It is also true that a number of censuses from developing countries present in the IPUMS-I database did not meet the criteria established here. These include Thailand 2000, Indonesia 1990, Peru 1993 and Ecuador 1990.

^10. In some countries (e.g. Turkey and Morocco) ‘children ever born’ of never married women were not recorded in the census data, making it impossible to estimate the non-marital fertility of these populations. Unfortunately, most censuses (particularly in developed countries) did not record the birth dates of children, so no reproductive timing measures can be estimated.

^11. However, according to 1990 Chinese Census data there was virtually no fertility boom among Chinese women, but an abrupt fall of fertility.

^12. In some countries, census data on marital status should be considered with precaution due to the well-known problem of infra-enumeration of single women (Esteve, García, & McCaa, 2010).

^13. It should be remembered that in Turkey and Morocco ‘children ever born’ variable is top-coded in the census data.

^14. Detailed country by country analysis is available upon request.

^15. These detailed data are not presented here but are available directly from the authors.

^16. There is no detailed information on education for Costa Rica or France.

^17. The decompositions follow the technique (‘components of a difference between two rates’ by means of ‘standardization with interactions allocated equally to the variables involved’) proposed by Kitagawa (1955, 1964) and applied, for instance, by Gibson (1976). France has not been included in this decomposition due to lack of data.

^18. Results of this decomposition exercise are available upon request.

^19. In Figure 5 only cohorts participating in the peak fertility boom years are shown (see Table 1), though the decomposition was carried out for birth cohorts prior to, during and after the boom itself. These data are also available from the authors upon request.

^20. According to cohort data presented by Frejka and Calot (2001a, 2001b), there was no baby boom in Finland, the former FRG, Portugal, Italy, Slovakia and Bulgaria either. Conversely, the Netherlands, Norway, Sweden, England and Wales, Australia, and New Zealand did experience baby booms.

^21. The glowing accounts of Tibor Mende of Latin America (1952) and Asia (1951) reflect this optimism, at least in the eyes of an expert international observer.

^22. Often called the ‘ski-jump’ effect of the historic transition, this brief and mostly modest turnaround in fertility took place during the third quarter of the nineteenth century in some but not all European nations (van de Walle, 1998).