Previous - Next
This is the old United Nations University website. Visit the new site at http://unu.edu
The demographic transition in Colombia
Background
The study
The
findings
Conclusions
References
Carmen Elisa Flórez and Elssy Bonilla
Centro de Estudios sobre Desarrollo Económico (CEDE),
Facultad de Economía,
Universidad de Los Andes, Bogotá
Over the last five decades Colombia has experienced a demographic transition, a change from high to low birth- and death-rates. It began towards the end of the 1930s when mortality rates started to decline. The crude mortality rate decreased from 30.5 to 9 per thousand, and the infant mortality rate from 200.2 to 61 per thousand, between 1938 and 1978. Life expectancy at birth is the average number of years a person can expect to live, calculated on current mortality rates for each age-group. This increased from 44 to 61 over the same period. Later, at the beginning of the 1960s, fertility started to decline sharply. The total fertility rate is the number of children a woman may be estimated to have on the basis of the current pattern and projections for those in her age. It decreased from 7.04 in 1960-1964 to 4.6 in 1972-1973 and to 3.6 in 1980, a reduction of almost 50 per cent in less than 20 years. The crude birth-rate dropped from 45.2 at the end of the 1950s to 28.9 per thousand in 1980. The decline began in the towns; although the decrease in rural areas was equally dramatic, there was a time-lag of almost 12 years before it began. In 1980, rural fertility levels were as high as they had been in urban areas in 1968, with a total fertility rate of 5 per cent. However, by the time of the 1985 census, the total fertility rate was 3.2 for the nation, 2.7 for urban areas and 4.6 for rural areas (Flórez, Echeverri, and Méndez, 1987). The combination of these trends resulted in what has been called a "demographic transition without precedent."
Much of this paper was written while the first author was a Hewlett Foundation post-doctoral fellow at the Population Research Center, University of Chicago. It is a summary of two studies, "The Impact of the Demographic Transition on Colombian Households" and "The Meaning of the Demographic Transition in Households of a Colombian Rural Setting," done at CEDE between 1983 and 1987 by a group in which Rafael Echeverri, Bernardo Guerero, Argemiro Morales, Martha Rodríguez, and Diana Medrano worked with the authors. Both studies were supported by the Household, Gender, and Age Programme of the United Nations University and the Female Status and Fertility Program of the Rockefeller Foundation.
During the same period, the country went through a number of important developmental changes. Without examining these socio-economic changes in detail, those which had the most significant effect on population structure were, first, the government's countrywide health campaigns oriented towards preventive medicine, which began around 1950, and, second, from 1956, a remarkable increase in public expenditure on education, which led to a decrease in illiteracy from 47.7 per cent in 1938 to 20.6 per cent in 1973, and a rise in elementary-school attendance from 56.2 per cent in 1951 to 89.1 per cent in 1973 However, the increased funding allocated to education and health was distributed unevenly and favoured urban areas. This factor, combined with the mechanization of agriculture in the 1950s and the growth of industry and construction in the 1960s and 1970s, caused unprecedented migration from rural areas. The percentage of the population living in urban areas doubled in 35 years from 30.7 per cent in 1937 to 62 per cent in 1937. Then, at the end of the 1960s, the private sector began providing family-planning services. This was permitted and tacitly supported by the government, thereby accelerating the fertility decrease which had begun earlier in the decade. Thus modernization and health and education policies combined to shape the demographic transition in Colombia. This process is expected to continue at least until the end of this century, although not at the same level of intensity as in the 1960s and 70s. However, analysis suggests that, even after the trends stabilize, large differences will persist between urban and rural areas, and between socio-economic sectors.
Demographic transition and falling fertility have a direct effect on the age distribution of the population. Colombia's population is becoming older; the relative proportion of infants is diminishing, while the proportion of adults and the elderly is increasing. This change has important implications at both the micro-and macro-levels. At the micro-level, it implies different types of family composition and different processes of family formation. At the macro-level, it means, amongst other things, different types of social demands on the state and different pressures in the job market. The demographic transition in Colombia has certainly had an enormous impact, as yet only partially analysed, on many spheres.
The speed with which the demographic changes occurred made possible the simultaneous examination of households in which women were of reproductive age when the total fertility rate was at its maximum level, and households where women were of reproductive age after the fertility decline. The principal objective of this study is to analyse the impact of Colombia's demographic transition at the micro-level, through a comparison of the life-courses of two groups of women: those who were at the peak of their fertility between 1960 and 1964, when fertility was at its highest (the 25-29 age-group at that time), and those who were at the peak of their fertility (between 20 and 24) in 1980, the most recent point for which relevant figures were available. It is thus possible to compare the lives of women before and after the demographic transition by examining differences between the two birth-cohorts (or groups of people born in the same period) at various stages of family formation and expansion. The quantitative data obtained were complemented by recording women's attitudes and perceptions to these processes.
Fig. 1. Colombian rural longitudinal survey.
Because of the considerable variation in incomes, the sample was divided into three socio-economic strata in both the urban and rural areas. Given the range of the study and the well-established differences in the timing and causes of the transition, two studies were set up one in an urban area, Bogota, the capital, and one in rural areas in the same Central Andean region.
This is the first longitudinal study of family formation as a dynamic process in Colombia. Most previous discussions of the demographic transition related the decreases in fertility and mortality to changes in the socio-economic and cultural variables associated with modernization. This study follows development through changes in the lives of individuals. Life-course histories charted with retrospective data permit a fuller understanding of the causes and processes of the transition in relation to family formation and develop a long-term historical perspective which produces a more dynamic analysis of change than that used in conventional demographic studies. This demands more than statistical methodology: it calls for a particular approach and research strategy which will inform both the collection and analysis of data.
Women are the primary focus of study. The household, which was defined as a co-residential unit whose members share at least one of the three daily meals, is approached through the woman. Women's behaviour and their perceptions of the household and related issues are the core of the analysis. Women's educational level and labour-force participation, as well as other variables like origin and place of residence, migration, and non-remunerated domestic and agricultural work, were included as factors in family formation and transitions. It would have been useful to include similar information about their partners, but collecting this would have added considerably to the cost of the study. Retrospective information was obtained from cohorts of women separated by an average of 20 years, so that the events recorded reflect a range of historical and contextual experiences. The analysis was based on categories which allowed a comparison of the two cohorts according to their socioeconomic position.
Both the urban and rural projects were divided into two parts, a longitudinal study and a qualitative and time-allocation study. The longitudinal study collected data on the present circumstances of the woman and her household and on her life-history. A life-course matrix, a simple way of relating personal and macro-level events and chronology which is easy to administer at any educational level, was used to do this. The life-history information provided the basis for a multivariate analysis of reproductive and productive behaviour over time. The qualitative and time-allocation study captured women's perceptions and roles as well as the division of labour within the household. This part of the survey was confined to a subsample from the longitudinal study because of the amount of time needed for the in-depth interviews.
Within the perspective of life-history, women's activities can be seen as behaviour that develops continually throughout their lives as they grow and age in their social environment. This social environment is characterized by experiences and opportunities associated with the community, economic status, position in the family, and the normative and biological restrictions on sexual and age behaviour. Women's education, marriage, labour-force participation, occupational status, and income define the context of their life-histories. Moreover, a life-course perspective permits an investigation of the factors in a woman's life which influenced the demographic events related to her family formation process - marriage and first childbearing, for example- at a specific point in time. Rather than following the usual practice of analysing a cumulative process (like children ever born), this study was able to analyse the occurrence of single events over transitions which affected reproduction (like marriage and fertility), as well as events spread over a life-span.
The retrospective longitudinal data made it possible to examine the determinants of such transitions. These data described sequences of events which occurred over a long period. However, the time covered does not include the complete life-histories of the women, which vary in length from 25 to 49 years, depending on the cohort. While each life-history reflects the events that occurred from birth to the time of the survey, the lives of the women of course continue after the interview. Some of the women were already married, had borne their first or second child and completed other demographic transitions by the time of the survey. Others had not yet completed any of the transitions, but might do so later: for example, a woman childless at the time of the survey could give birth subsequently. The experiences of these women are censored by the date of the survey. This problem applies particularly to those in the younger cohort whose experiences are thus cut off at a relatively early age (25-31 years). However, the proportional hazards life-table models developed by Cox (1972) and others were used to overcome this bias in the analysis of longitudinal data.
Ethno-methodology and the time-allocation survey complemented the longitudinal perspective. The ethno-methodological approach was designed to depict individuals and their social interactions from the standpoint of their own interpretation of themselves and their activities. This can enhance understanding of women's perceptions of the factors that influence the productive and reproductive decisions reflected in their life-history data, and help reveal attitudinal and behavioural trends in the different age-groups. The study also sought to identify the conceptual framework underlying the women's interpretations. The analysis was enriched by the continual interplay of the qualitative and empirical material, which enabled the study to capture general cultural patterns in the various strata and cohorts and to offer broader substantive interpretation.
As both women's attitudes and perceptions and the distribution of household and productive activities changed through time as a consequence of socioeconomic modifications and the demographic transition, the study made use of demographic, statistical, and sociological methods of analysis. Conventional demographic measurements such as nuptiality and fertility patterns and indices of birth-control usage established the differentials between the cohorts in purely demographic terms. Nuptiality and fertility models developed by Coale (Coale, 1971; Coale and Trussell, 1974) were used to identify the principal phases of family formation and expansion. These methods, however, did not allow for the full exploitation of the available longitudinal data. Specifically, they did not examine the determinants of transitions between statuses, consider the effects of various independent variables on the likelihood of demographic events, or interrelate them at the time they occurred. Therefore statistical techniques especially designed for multivariate analysis of life-history data (Tuma et al., 1979; Trussell and Hammerslough, 1983) were also used.
The urban survey was conducted in the capital, Bogota. The sample consisted of women of 45 to 49 and 25 to 29 living there in private houses between May and June 1984. These two cohorts respectively represent behaviour before and after the demographic transition. The sample was stratified according to an inventory of dwelling units in the city compiled by the National Administrative Department of Statistics (DANE) in 1980-1981, which classified the dwellings by external observation of the type of construction, quality of the unit, availability of utilities, location, and related data.
The target urban sample number of 1,500 women (750 in each cohort) led to visits to 6,756 dwellings in 360 clusters (120 in each stratum) from which it was possible to get information from 6,012, a coverage of 93 per cent. In these selected units there were 8,430 households, including 630 women in the younger cohort and 623 in the older cohort. Of these women, 91.7 per cent of the younger cohort and 79.5 per cent of the older cohort agreed to be interviewed, giving respective totals of 578 and 496 completed interviews.
The rural survey was conducted as a basis for comparison with the Bogota study. For this reason, and in order to control for differentials resulting from different cultural patterns, the rural areas chosen shared certain geographical and cultural features with the capital. Analysis of migration flows indicated that the states of Boyacá and Cundinamarca accounted for almost 80 per cent of the migrants to Bogota. The study therefore concentrated on townships of less than 1,000 people which were covered by the government Rural Integrated Development Program in outlying areas of these two states. They were all located in the higher part of the Andean region relatively close to Bogota, at an altitude similar to that of the capital.
The rural sample consisted of women of 40-49 and 25-31 living in private households in this area between October and November 1986. The age-range of the younger cohort was extended further than in the urban sample to avoid statistical problems related to the different numbers of women in the relevant populations in the two areas. Lack of information precluded a pre-stratified sample. The target sample of 600 women in each cohort led to a selection of 3,600 dwellings in 360 clusters. This entailed visits to 3,537 units, 3,284 of which were occupied (7.1 per cent vacancy rate). It was possible to get data on 3,059 dwellings (93 per cent coverage). These included 3,369 households, 3,133 of which gave information (93 per cent coverage). From these households' 570 women were selected for the younger cohort and 638 for the older cohort. Five hundred and thirty-three women in the younger group and 578 in the older group agreed to be interviewed (93.5 and 90.6 per cent coverage respectively).
In order to control for socio-economic conditions, the townships, the households, and the women were stratified ex post facto. Three socio-economic strata were defined according to the state of the household dwelling, conditions of tenure, economic base, isolation, access to services, and community organization. The women were then classified into socio-economic strata according to the household to which they belonged. Those who lived in the most precarious conditions were in the lowest stratum. Those who enjoyed the highest living standards were assigned to the top stratum, and the remaining third to the middle stratum.
Both the urban and rural areas studied had better than the national average access to education, health services, roads, water supply, sewerage, and electricity. However, within each zone, differences between the socio-economic strata were highlighted by the state of the basic infrastructure and the quality of the dwellings.
Although the levels of education in Bogota were high in relation to the rest of the country, access to the higher levels was poor. There was good access to primary education in the rural zones, but retention rates were low; there was very little access to secondary education, with relatively low levels of educational attainment among the population observed.
Bogotá is a metropolitan centre with a large tertiary economic sector. Patterns of participation in non-paid family agricultural activity prevailed in the rural areas, and household resources were derived mainly from a combination of this and work for subsistence and wages, though there were notable differences between the strata.
Over the last 30 years there has been considerable immigration into Bogota from other parts of the country and not many of the urban women studied (17 per cent of the older and 40 per cent of the younger cohort) were native to the city. The migrant women, especially those in the lower strata, were of predominantly rural origin. On the other hand, the majority of the rural women interviewed (at least 90 per cent in each cohort), were native to the area, and a high proportion of them still lived where they were born.
Both the urban and rural studies used the same basic questionnaire. This was designed to gather two sets of information, one on the woman and the characteristics of her household at the time of the interview, and the other on her life-history. The first part included information about members of the household, including kinship, age, sex, marital status, and participation in wage-earning activity and housework. It also recorded facts about the woman's parents - their education and work when she was 15 years old - and her own life at the time of the interview. The second part of the questionnaire sought information about the sequence of the events of the woman's life.
The rural study used two additional instruments for cluster and household analyses. The cluster questionnaire gathered the information necessary for the retrospective stratification, including data on altitude, isolation, and community services. The household questionnaire gathered information on both the dwelling unit (typology and availability of basic infrastructure) and the household itself (land ownership, main resource base).
The qualitative part of the study sought to identify the changes between the cohorts, which represented behaviour before and after the demographic transition, in order to arrive at a better understanding of the observed differences in their family formation processes. Two different instruments were used with a subsample of 30 women, 15 from each cohort. The first, a highly detailed guideline for the in-depth interviews, was designed to elicit women's ideas, opinions, and attitudes on issues such as marriage, sexuality, motherhood, abortion, family planning, work, power relationships, and female participation in various social contexts. In view of the significance of female labour-force participation in the longitudinal study, it was considered particularly important to capture the women's perception of waged work. Secondly, for a subgroup of 10 women from each area, a time-allocation chart was designed to map their domestic work, whether paid or unpaid, and child-care and leisure activities. This showed the distribution of work among the different members of the household. In rural areas the interviewers used a video camera to record more detailed information about the environment and life-styles of a subsample of the women.
Family Formation and Expansion
The women's marital status at the time of the interview suggested marriage patterns that differed between urban and rural areas and between socio-economic strata within each area. These variations certainly influenced fertility. The proportion of women who had ever used contraception was significantly higher among the younger women. The impact of this on fertility became stronger with the growing use of more effective and modern methods like the pill and the IUD. The increased practice of birth control in both urban and rural areas was a consequence of both higher levels of education, which led to greater acceptance, and better coverage by family-planning programmes, especially in the country. Rural women in the low stratum were the least likely to use contraception, and they also had the lowest levels of education and the poorest access to family-planning programmes.
The relationship between female productive and reproductive behaviour is complex and mutually responsive. Female labour-force participation affects family formation and marital status and vice versa. Marital status was the variable with the most significant and strongest impact on the probability of working. The number of children also had a negative effects, which highlighted the incompatibility of productive and reproductive roles among both urban and rural women. Female labour-force participation increased in all strata from the older to the younger cohort in both urban and rural areas. A concomitant decrease in fertility was noticeable between the younger and the older urban and rural cohorts. Clear patterns in the data confirm that having a family diminishes female labour-force participation. However, the constraining effect of children appears to relate more strongly to their age than to their number. Conversely, the number of children a woman had was restrained by labour-force participation: those with waged-work experience bore fewer children than those who had never worked. The data also suggest that the increased opportunity costs of women's time as a consequence of substantial improvement in their education has led to higher workforce participation. In line with this, there has been a movement towards more qualified occupations in the city and away from family agricultural activities in the rural area. These interrelationships were examined in several ways through multivariate analysis.
The completed histories of fertility and nuptiality allowed the comparison of the processes of family formation and suggested that there have been substantial changes in the formation phase, as well as in the expansion of families between the generations, again with great differences between urban and rural areas and between socio-economic strata.
The data show that the most dramatic changes in patterns of first marriage occurred among urban women in the upper and, to a lesser extent, the middle stratum. Without a comparison of the economic strata it could have been erroneously concluded that no major changes had occurred in the family formation phase: the urban age at first marriage (which was taken to include both legal and consensual unions) was around 21.5 years and the proportion of women ever married was about 91 per cent for both cohorts. For rural women, the figures were 20.4 years and 87 per cent respectively. However, behind these similarities between the cohorts at the aggregate level, there were big differences between the urban economic strata. In the top stratum, there was no significant increase in the mean age at marriage (which only moved from 22.1 to 22.6), but at each age the proportion of marriages fell, producing a drop in the percentage ever married at the age of 50 from 85 to 62 per cent. Contrary to the marked changes in the upper urban stratum, the pattern in all the rural and the lower urban strata remained fairly stable. This differential behaviour is of great importance as the total fertility pattern is a product of two patterns: the nuptiality or first marriage pattern, and the marital fertility pattern, so that a decline in the total fertility of upper-stratum urban women could be expected to follow the new trends in their marriage patterns.
The data show increased use of family planning and a sharp decline in both urban and rural fertility from the older to the younger cohorts. The total fertility rate decreased from 4.7 to 2.6 from the to the younger urban cohorts, and from 6.0 to 3.4 in the rural cohorts. Although the data on family planning indicates more use of contraception in all strata in both areas, the greatest decline in total fertility occurred among urban women in the upper stratum, presumably partly due to the drop in their first marriage rates. On the other hand, the total fertility rate of urban women in the middle and lower strata and of all rural women decreased without significant changes in nuptiality. Thus, although falling marriage rates probably contributed to the lower fertility of urban women in the upper stratum, nuptiality has not played an important role in the overall fertility decline.
The pace at which the urban cohorts expanded their families show that while there have been major changes in the spacing between children, the interval between marriage and the birth of the first child has not altered substantially. This suggests that the younger urban cohort used contraception to time the births of their children and to have a smaller family. The rural data, on the other hand, show that the smaller families of the younger cohort there were achieved without major changes in the former patterns of timing. This implies that the fertility decline in the country areas has mainly come about through contraception after the desired family size has been reached. This result supports the findings of the qualitative study, which indicated that rural women favoured birth control only among married women and after they have had some children. Thus both the methodologies used suggest an important difference between the urban and rural areas in the use of birth control in family formation.
Women's educational attainments and work experience had a marked impact upon the beginning and expansion of their families. The likelihood of a woman having a third child in relation to her education and work experience was examined by analysis which included origin and place of residence as socioeconomic variables. The point at which education begins to have a strong effect on patterns of first marriage and the birth of the first child was the completion of elementary schooling. It seems clear that the higher educational levels of the younger urban cohort have been important factors in the greater changes in their nuptiality and fertility.