The Determinants of Brazil's Recent Rapid Decline in Fertility (1983)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

PART I ~ FERTILITY DETE^IN - TS AT T~ ~ "WL . CHAPTER 5 WE NIlIR: PURPOSE AND METHODOLOGY As noted in the Introduction, the discussion in this and the following chapters parallels Part I, but with a local- rather than a national-level focus. There are, of course, advantages to be gained from both: the national- level analys is permits aggregate conclusions and hypotheses, however tentative; the local-level data examined in the following pages provides another , complementary per specti~re on those perceptional. These chapters examine the purpose and methodology of the NIlIRs levels and trends in the total fertility rate for the nine local contexts studied; and the effects of nuts tiality, marital fertility patterns, the proximate determinants, and socioeconomic factors, specif ically income, on fertility for the nine contexts. The final chapter shows that the national-level and the local-level conclusions are consistent. The NIBR was conducted between May 1975 and March 1977 by Centro Basileiro de Analise e Planejamento (CEBIW?) in Sao Paula, under the f inancial auspices of the Inter- national Development Research Centre (IDRC, Canada), the Population Council (USA), and Financiadora de Estudos e Projecto'; (REP, Brazil). It consists of a series of local case. studies based on a view of population dynamics as part of the overall structural dynamics of society, with economic determinants (manpower requirements, forms and degrees of accumulation, property patterns, etc.) having a decisive influence. However, this dominance of socioeconomic factors over population dynamics does not operate in an obvious way. Therefore, rather than simply assuming such relations, it is essential to examine analytically and empir ically how they are established. Although structures condition behavior, and behavior only exists through the actions of individuals With their 145

146 aspirations, motivations, forms of personality, Audi visually calculated interests, etc. ), the mediation between structures and action is essential. If the - mediations are not taken into account, the analysis TaUSt inevitably become static. The result is the deduction of r igid prescriptions for action from structural patterns, or, at the opposite extreme, the dissolution of social conditioning factors into the subjectivity of the individual agents, so that the only way to understand human reproduction Is to take a purely psychological approach. The NIHR perspective takes the role of social institutions the Church, private enterprise' the educational system, the medical system, and so on- as being crucial to such mediation. The way population practices and values are actually realized, and population policies defined, depends on how population is seen in the context of these institutions, and how, within each one, interests and values concerning the family, number of children, procreation, and family planning are articulated. The NICER began by choosing the areas of Brazil that would represent as purely as possible the different forms in which the production is organized. A two-d`mensional typology of the areas was then developed according to their dominant forms of production and their place in the process of development.2 In the former category' three modes were distinguished, each of which might predominate in a given area, although in combination with the others: the capitalist mode, a system of small autonomous pro- ducers, and rural servitude. The capitalist form was subdivided into monopolistic and competitive; for the small autonomous producers, the distinction was between rural and urban. The second dimension, place in the pro- cess of development' was based on the assumption that development tends to concentrate economic activities and population in certain areas, and to disperse them in others. The former were classed as reconcentration areas,. which are invariably urban; the latter were labeled ~dis- Persian areas,. which as a rule comprise extractive or agricultural activities, with the formation of frontiers at the edges of an overexpanding transportation system. This classification also left room for certain areas, labeled interstitial, which fall between the types des- cr~bed above, and which tend to lose activities and popu- lation; these areas may be either urban or rural. These two dimensions could produce a total of 15 pos- sible types. However, four of these were discarded as

147 meaningless: the monopolistic and interstitial capitalist area; simple commodity production with an urban location in a dispersion area; simple commodity production with a rural location in a concentration area; and rural servi- tude in a concentration area. The monopolistic capitalist area undergoing dispersion was also discarded since it was considered highly rare in its pure state in Brazil; more- over, within a given gradient, it might appear in some regions classified as comuetitive-canitalist in a dis- persed form. - ~~ Finally, rural servitude in an interstitial form was also omitted since this combination might be found, at least partially, in regions defined as sample commodity production with a rural location and intersti- tial form. The next stage of analysis consisted of making the nine remaining possibilities correspond to actual regions in Brazil. Regions were distributed both spatially and, whenever possible, according to the dual classification. The results are shown in Table 50; the contexts selected are described in detail in the Appendix. The nine areas chosen cover seven municipalities {for the sake of convenience, the rural and urban categor les in the sys- tem of autonomous producers are represented by the rural and urban areas of the same municipalities, thus yielding nine area types and only seven municipalities). These municipalities are distributed over six states: Sao Paula, Rio Grande do Sul, Pernambuco, Espirito Santa, Para, and Pisui. The map of Brazil (see page 14) gives an idea of this geographical distribution. For each of the nine chosen areas, two types of study were performed: a macro-structural study covering the whole population of the area, and following this, a survey based on a probabilistic sample of the population in the same area. The arm of the macro-structural study was to acquire as complete as possible a knowledge of the dynamics of the population, its socioeconomic structure, and the area9 s social institutions from its foundation to the present. The three components of population dynamics considered were nasality, mortality, and migration. The socioeco- comic factors considered essential to "~he study's objec- tives were the area's landowning structure, product mix, and predominant types of activity; the number of persons engaged in work and He types of occupations (present and previous); the duration and type of land occupation; prom auction values and forms of appropriation of surplus; and production techniques. The study of institutions focused

'48 TABLE 50 Nine Contexts Defined in Terms of Modes of Production and Type of Involvement in the Development Process: Brazil Insertion in Development Process Modes of Production Concentration Dispersion Interstitial Capitalism Monopolistic Sao Jose dos Carom SSso Paulo - SP) __ Competitive Recife Sertaozinho Cachoeiro do (Pern~uco - - (Sao Paula - I~apemirim PE3 SP) (Espirito Santo - ES) System of Autonomous Producer 5 Urban Santa Cruz do -- Parnaiba Sul (Rio Grande (Pisui - PI) do Sul - RS ) Rural -- Santa Cruz do Parnaiba Sul (Rio Grande (Pisui - PI) do Sul - RS ) Rural Servitude Conceicso do -- Araguaia (Pare - PA) on their role in individual reproductive decision making. The institutions considered included the economic produc- tion unit, the family, the health care system, the educa- tional system, religious and political institutions, and the mass media. The sources of information used were secondary: censuses (demographic, industrial, and agr'- cultural); other specialized publications containing statistical data; historical monographs; papers on research carried out previously in the same region; direct observation with qualified informants; focused interviews with people who had lived in the region for a long time; and in~depth intprUi - I "i ~ h m~mh^'c ^F The ~1=~;~" ; n generals 3 ~—- - - - -~ r—Rae—~ ~ ^~^ -as The second stage of the NIBR consisted of a fertility survey conducted in each of the nine areas, shaped by the findings of the macro-structural study. That initial

149 study helped delineate the sample by revealing, for example, the production relations prevailing in the area ; the rural proletariat and its spatial distribution, both In the countryside and on the urban outskirts of a munici- pality; the annual cycle of each region' s range of ma jor products, coupled with the rainy and dry seasons; and the seasonality of harvests requiring harvest workers. In- depth interviews among the population in general were another important aid in designing the survey. The sam- pling procedure proposed for the survey was three-stage equiprobability: the primary sampling unit was a census tract, the secondary unit a block, and the tertiary unit a household. ~~ -~ whenever possible, appropriate. sampling tractions were determined to ensure that each household in a given region had the same probability of belonging to the sample. A household was understood as a group of people linked by kinship, affinity, or economic relations and living under the same roof. Each was given a questionnaire con- sisting of three modules. The first, known as the household module, contained information on all the members of the household, identify- ing them through their relation to the head of the house- hold, as well as their age, sex, marital status, level of schooling, occupation (type of activity, position, and periodicity), monthly earnings, and consumption. In addition, there was information on help given to or received from other members of the group not living in the household, and on the division of labor within it. In rural areas, additional questions concerned the different arrangements for land tenure, types of harvest, number of animals, agricultural Implements, and the like. During the interview, this information was given by any adult considered a qualified informant and a member of the household group. The second module, a life history, was designed to obtain fundamental quantitative data for a study of repro- duction strategies. Unlike most fertility studies, the NIBS was not limited to women between ages 15 and 49 in stable marital unions.4 Any member of the household group was eligible for this module, as well as for the third, described below.5 Some clarification is required concerning the inclusion of males in this survey. Although rarely carried out in practice, the need for this inclusion has been pointed to with some frequency in the literature since the husband's role in decision making within the family, his position on the use of contracep-

150 fives, and so on are important determinants of reproduc- tive behavior.6 In the NIER, inclusion of males is justified not only for these reasons, but also for others resulting from the theoretical approach adopted: since the NIHR set out to link the forms of production to modes of reproduction, the agent most closely linked with the production process was especially important. One indivi dual was selected at random from each household for application of the life history module and referred to thereafter as the Ego. The Ego's life history was fol- lowed from birth to the present. Information was gathered on migration, education, and occupation, as well as on the composition of and changes in the Ego' s family group e The module also contained a detailed reproductive history, beginning with the menarche ( for women) or earliest sexual relation {for men) and continuing up to the menopause or andropause' respectively, with information gathered on unions, duration and result of pregnancies, use of con- traceptives, breastfeeding, surviving children, dates of death of nonsurvivors, and so on. If the Ego then had previous or current marital unions, the life history of the spouse was gathered in equal detail f rom the beg inning of the union. Information given by Ego was also used to reconstruct, in less detail' the spouse's life history at three points prior to the ur~ion--at birth' age 10, and age 18. Finally, the third module, also answered by the Ego, was aimed at obtaining transversal material to explain a number of actual practices involving the relationship between institutions and reproductive behavior. These included the existence of distinct plans or strategies regarding reproduction, help given to and received frown children, and the like. In conclusion, it should be emphasized that the NICER was not intended as a representative national sample. Its samples are representative only in relation to the area or municipality from which they were selected. Given the characteristics of the study, it would be practically meaningless to aggregate the nine contexts to provide estimates of national~level demographic parameters. Thus the discussion that follows, while parallel to that of Part ~ in many ways, is quite different in focus, and serves as a complement to that analysis. -

CHAPTER 6 THE TOTAL FERTI LI ~ lo: AWLS AD TRENDS Before proceeding with an analysis of the total fertility rate based on NIHR data, it is worth clarifying some points with regard to these data. The NIBR surveys were performed with the aim of covering 400 households in each area except Sao Jose dos Campos and Recife, where 800 households per area were included. As mentioned above, the information unit was any adult. However, for purposes of the following analysis, the unit will be taken as a woman, and some comments should be made in this con- nection. First, it is worth noting that the information on a worn ' ~ reproductive history {contained in the life- history module) was given by the woman when she was the selected adult, or by her mate when he was selected, whether his wife was allure or not. Thus, the samples do not refer only to cohorts of women who survived until the survey, and it is not necessary to make the usual assump- tion that the fertility experience of women who did not survive is faithfully represented by the experience of those who did. With this in mind, whenever the rep~oduc- tive history of an adult is considered, the woman may have been in any one of the following situations: 1. alive and single 2. alive and married 3. alive and separated 4. alive and widowed 5. dead and married while alive Situations 1, 3, and 4 could only occur if the woman was the informant; situation 5 could only occur if a man was the informant; and situation 2 could occur in either case. Second, the fact that the investigation also included single women avoided the problem of completeness involved 151

152 in leaving out all childbearing experienced by women who declared they were single at the time of the survey. At the same time, it meant that, in general, the premarital experience of women ever married at the time of the survey was not eliminated. Third, it should be stressed that in this investigation there was no cut-of$ age for asking questions on fertility history. Thus the data were not truncated, as usually happens when histories are not obtained for women over age 50. This means that fertility rates could be cal- culated for almost all age groups and for a distant past, although it must be remembered that memory problems man introduce certain difficulties. The total fertility rates (Table 51) were established for three different points in time: 1965, 1970, and 1975. For each of these periods, the rates were calculated on the basis of the total number of women alive at that time, regardless of marital status. Thus if a woman was alive, for instance, in 1965, she was not necessarily alive in 1970 or after. Moreover, when the informant (Ego) was a husband separated from his wife during the year for which the fertility rate was calculated, this case was el~mi- nated, since there was no way of knowing that woman's situation with regard to ache number of children between the year of separation and the year of reference for cal- culating fertility. A further clarification is required concerning the dates of the surveys. For seven out of the nine areas, the surveys were conducted in 1976 Or 1977; thus, the data for births in 1975 present no prob- lems as to the cat endar year. However, in Sao Jose dos Campos, the survey was performed from May to December 1975, and in Santa Crud do Su}-Urban from November 1975 to July 1976. ~ correction was thus necessary for these two areas since the women in both were not exposed for the whole year. For Sao Jose dos C ~ os, the number of births in 197S was multiplied by the factor 1.53G, and for Santa Cruz do Sul-Urban by 1.04S. The first observation to be made when analyzing Table 51 is that there is some consistency between these results and those of Table 2 from Part ~ of this reports Indeed' the 1970 rate for Sao Jone (4.83) was consistent with that for Sao Paulo State (4.07), especially given that more than 70 percent of the latter's population was urban in 1970. Similarly, for Parnaiba-Rural, Parnaiba-Urban, and Recife, located in Northeast Brazil, total fertility as reported in Table 51 varied from 8.42 to S.45 in 1970; the Table 2 value for the Northeast (7.58) is within this it, , ~ ,

153 TABLE 51 Total Fertility Rates, Nine Contexts, 1965, 1970, and 1975: Brazil Total Fertility Bate Percent Decline 1970. 19?s. 1975~. Context 1965 1970 1975 1965 1970 1965 Parnalba-Rura1 `82;l)2 ~2S3)2 (258) S.94 +1202S +SeS8 Sso Jose dos Cantos (4i3) (534) (S92) 17.20 16.56 30.87 Recife (405) (503)8 (45;13° 1.28 8.92 10.09 Concsiceo do Araguata (201)8 (2;2)6 (36i49)6 +1.11 11.02 10.03 Parnaiba~rban (2il) (2;2)6 (6289) 5.48 21.S2 25.82 Sertaozinho (2i7) (2;6)7 l3;6s)8 1.22 35.10 3S.89 Cachoeiro (20864)8 (261~8 (230917 2S.64 14.66 36.~d Santa Cruz-~ral (2il) (2;3} (2;8) 16.62 2S.53 37.91 Sen" Cruz~Ur~ban (2;3) (2;1) 2;92 41.61 +14.96 32.87 Notes N~re in p"entbe - s are number of won. range. It should also be noted that in 1970, 58 percent of the northeastern population lived in rural areas. Finally, for Concaicso do Araquaia, a frontier area, Table 51 shows a total fertility rate of 7.26 for the same per- iod; this is very similar to the Table 2 rate for Frontier States of 7eG8. There was a very large variation- in fertility from one context to another, and thin variation increased over time. In 196S, the maximum variability was about 4 chil- dren per woman, rising to 5.4 in 1970 and 6.0 In 197S. The extreme cases were always Santa Cruz do Sul-Urban, with the lowest total fertility rate, and Parnaiba-Rural, with the highest. Figure 5 (on page 37) highlights the enormous differences in fertility between these two

154 contexts: the reproductive pattern of Santa Cruz-Urban is more like that of a European country with low fertility, while that of Parnaiba-~ural resembles a natural-fertility pattern. ~ further point, which also confirms the theoretical approach of the NIHR, is that the rural contexts do not necessarily have higher fertility rates than the urban contexts. In analyzing the trends for each context, it can be seen that from 1965 to 1970, with the exception of Con- ceicao do Araguaia, all areas show a decline. This decline was most marked in Santa Cruz-Urban (42 percent), Cachoeiro (26 percent), Sao Jose dos Campos (17 percent), and Santa Cruz-Rural (17 percent); Parnaiba-Urban and oRural had a slight decline of around 5 to 6 percent; and Recife and Sertaozinho had a fall of only 1.2 percent. For 1970 to 1975, the decline was much steeper, except for Parnaiba-Rural and Santa Cruz-Urban. However, the total fertility rate for the latter area in 1975 should be taken with reservations, since fluctuations in the sampling process made the specific fertility rate for women between 15 and 19 very high in 1915. For the same period, the percentage decline was consistently high, ranging from 9 percent to 3S percent; this explains the acceleration in Brazil's fertility decline of an average 24 percent. The decline in contexts located in North and Northeast Brazil ----Recife, Parnaiba-Urban and -Rural' and Conceicso do Araguaia ~ accelerated over the 1970-75 period, rising from 9 percent to 22 percent. In 1970-75, Sao Jose dos Campos maintained the same level of decline as for 1965- 7 Around 17 percent. Cachoeiro showed deceleration in 1970075 with a decline of only 15 percent, as compared to its 26 percent decline for the preceding period. For the lo years front 1965 to 197S, it is very clear that there was a fall of over 30 percent in Sao Jose, Sertaozinho, Cachoe~ro, and Santa Cruz-Urban and -Rural; Recife and Conceicaco do Araguaia maintained a fall of 10 pervert; and only Parnaiba-Rural showed markedly high rates. Another approach to changing fertility levels from 1965 to l9?S is to examine parity for women in the different contexts. Table 52 shows some of this information. To reduce the fluctuation in the sample due to the small num- ber of cases, only average parity based on that of the 25-29 and 30-34 age brackets was calculated, with the aim of ref lecting the average number of children up to 30 years of age. The drop for 1970-7S was also fairly steep--around 17 percent--for most areas, thus corroborat- ing earlier f Endings .

155 TABLE 52 Mean Parity 1/2 (P25-29 + P30-34), 1965, 1970, and 1975: Brazil 1970. 1975 Context 196S 1970 1975 1965 1970 Santa Cruz-Urban 2032 1099 1.65 14.22 1,.08 Santa Cruz-Etural 2.92 3.16 2.80 +8022 11.39 Cachoeiro 2.99 2.57 2.09 14 ~ 05 18. 68 Sao Jose dos Campos 3 o 13 3.02 20 48 3 0 51 17. 88 Recife 3. 27 3.09 2 0 72 S O 50 11.97 Sertaozinho 3.54 3.47 2. BE 1. 98 16 o 14 Conceicso do Araguaia 3.93 4025 4.34 +8.14 +2.12 Parnaiba-Urban 4.44 4033 3.58 2.48 17.32 Parnaiba-Rural 4 0 68 4 0 70 4 ~ 42 +0. 43 5.96 The central question posed by the data reported above is why cortex" with very different social and economic structures show such similar rates of decline, while more similar contexts show such different rates. The d~scus- sion that follows explores this question by examining the role of nuptiality and marital fertility in the fertility declines observed at the local level (Chapters 7 and 8); the role of the proximate determinants (Chapter 9); and the role of socioeconomic factors, specifically family income (Chapter 10).

CHAPTER 7 NUPTIALITY As will be seen in the next chapter, a decline in marital fertility was the primary determinant of the fertility declines noted above. This conclusion parallels that of the national-level analysis in Part I. Before analyzing marital fertility, however, it is important to examine some other determinants related to nuptiality -the type of union and the initial, mean, and final age at mar- riage--that can contribute to a rise or fall in fertility. TYPE OF UNION Distribution of Types of Union Table 53 shows the distribution of ever-marr~ed women from the NIER according to type of union in three periods of timeool960, 1910 f and 1915. Before this table is anal- yzed, some points must be made about the various types of unions During the colonial and imperial periods in Bra- zil, marriages were performed by the Catholic Church. In 189G, however, civil marriage was legally instituted as the only valid form of marriage. Since then, those who follow some kind of religion have continued to conduct religious marriages, either exclusively or as a complement to legal marriage. Two other types of union are present in Brazilian society: consensual union and permanent union. Both of these are also known as free unions since they are not limited by any type of binding connection, civil or religious. Both consensual and permanent unions are defined as stable: in the former, the couple live in the same house; in the latter, they do not. Comparing the nine NIER contexts tTable 53), for a fixed time period, say 197S, there is a great difference 156

157 in the types of union, ranging from the situate on in which practically all unions were legal (Santa Cruz do Sul- Urban) to the opposite extreme, in which less than 30 percent of couples legalized their unions (Parnaiba- Rural). It is also noteworthy, in the same line of thought, that in Parnaiba-Urban and -Rural, and in Con- ceicso do Araguaia, the category Religious Only accounted for a large nu~nher of unions; in the other contexts, free union was more common than religious union, even reaching one-third of all unions in Recife. This diversity mung contexts found for 1975 can also be seen for earlier per- iods from at least 1960 onward, as the data in Table 53 clearly show. Indeed, in 1960 the proportion of legal unions varied from 29.9 percent in Parnaiba-Rural to 92.5 percent in Santa Cruz-Urban; between these extremes, there were intermediate figures such as 52.0 percent in Parnaiba-Urban, 63 percent in Araguaia, and 66.6 percent in Recife. What is noteworthy here, however, is that for all contexts, the main change over the 15 years analyzed consisted of a gradual decline in religious as compared with free unions. In certain contexts, this decline was smaller, as in Sao Jose and Santa Cruz-Urban; in others it was greater, as in Parnaiba-Urban and -Rural, and in Recife. In Sertaozinho, the fall in the number of relig- ious unions was not sufficient to account for the increase found For free unions, but was partly due to a reduction in the relative weight of legal unions; the same can be said of Conceicso do Araguaia. This relative increase in free unions as a result of a fall in either religious or legalized unions is also shown in Table 54, based on the findings of the 1960 and 1970 censuses and the 1978 PNAD (National Household Sample Survey). It should be observed that in these sources, free unions include only consensual unions; there is no explicit mention of permanent unions. From this it can be concluded that the latter have been incorporated in either the consensual or single category. Indeed, the 1970 census defined single women as follows: Those women who have not entered into any civil or religious, or civil and religious marriage, and do not live in a stable con- sensual union.. The data in Table 54 also lend some validity to the findings from the NIER shown in Table 53 (which are depen- dent on sample fluctuations given the small number of cases): there is a great deal of similarity between the contexts (Table 53) within a particular region and the results for that same region (Table 54); in other words,

158 x o C) .,' At o .-, a: ~4 o P. .~ n 8 : o o · - TIC dJ N ~ Al · - ~ al he ~ US V of r" O TIC o - ::, O ~ ~ : O ~ ~ En Z O 4 A C) Al S O ~ w o - _ I: O `: :. V C a' o ~ e a: 0 ret ~ us ~ c' ~ ~ ~ ~ co ret fir ~ ~ ~ ~ ~ ID 0\ Up GO . · ~ ~ a' 0 lo ~ ~ cat ~ ~ a ~ ~ ~ ~ a ~ ~ ~ ~ ~ us ~ ~ r cry ~ ~ ~ A _4 ~ a' t" C9 ~ 11~ _O ~ ~P o o o · · o o o · o C o ~o ~ c~ ~ ~o ~o ~ a, O O ~ ~ ~ ~ ~ o o ~ ~ ~ o ~ ~ ~ ~ ~ O O O O 0` ~ ~ ~ C' O O . . · . . · . . · o . o ~ ~ ~ N ~ ~ ~ ~ O ~ ~ ~— a~ eo ~ ~ ~ ~ ~ ~ ~ ~ (D (D C~ 0 ~r ~ ~ ~ · . · · . ~ r~ ~ ~ ~ ,*' c~ ~ ~ r~ ~ ~ ~ o ~ · ~ ~ ~ ~ C~ ~ U~ ~ ~ E ~ E C) ~Q ~ O a: 0 1 ~ 1 ~ 0 3 c) ~ ~ ~ X O O 0m ~ O 0 - 0 0 - ~ O 0 - `0 ~ ~ ~ `0 mm ~ ~ mm .. C. mm =m ~C~ S: O ~ ~ ~ ~ ~ ~ ~ U ~ ~ ~ C O ~ ~ ts le U) ~ U U]

159 In ~ ~ ~ ~ (D O O to al to (D Cal ~ ~ ~ Cal fir al ~ ~ ~ 0 0 · ~ ~ e ~ \0 ~ ~ Am · 0 ~ 0 0 ~ ~ ret 0 ~ ~4 ~ - 4 (D ~ ~ ~ ~ Cal ret O · . ~ ~ ~ · · ~ ~ ~ A? ~ ~ ~ UP O ~ ~ ~ \0 `0 O ~ O ~ ~ \0 ~ ~ Cal 0 ~ ~ ~ ~ · e ~ ~ Us O ~ ~ Cal Cal O ~ O UP · 0 · · 0 ~ 0 ~ ~ `0 ~ ~ ~ ~ ~ `00 ~ ~ ~ Cal · 0 0 O O O `0 - m 0 ~ ~ ~ u, O ~ ~ ~ O ~ en ~ ~ ~ ° e ° a Id Cal ~ 0` ID O ~ o up ~ up ~ ~ ~ ~ ~ ~ ~ us ~0 ~8 ~0 Set - - ~ _1 - 0 Cal ~ ~ 050 ~ mm · ~ 0 · ~ · · e · ~ ~ · ~ · In ~ UP us ~ ~ Cal ~ en us ~ ~ ~ In ~ mm In upon ~ mm I: :: _ O 0= t~ ~ t—t— C ~ CP, ~ _I _~ _ a' ~S 3 ~: 1 D _~ o o ~ C~ o S ~: O O O ~ ~ - , cn L' o o C) O O V ~ m01 C ~ - - ~ O 0 - tW ~o w

160 TABLE 54 Percent Distribution of Ever-Married Women, by Type of Mar ital Union, 1960, 1970, and 1978: Brazil Type of Union Con text Religious Civil and Religious Only and Civil Only Consensual Brazil 1960 20 O 2 73. ~ 6. i970 140. 78~6 801 1978 8 O 1 81 O 2 10.7 . Sao Paulo 1960 5~1 9201 2~8 1970 301 92~8 4~1 1978 1e 1 90~3 8eO Southern States 1960 9.6 86.8 3.6 1910 6.2 89.7 4.l 1978 3.5 90.0 6.5 Minas Gerale/ Esplrito Sento 1960 17.9 78. a 3. 2 1970 lO.B 85.2 4.0 1918 5. 0 89.0 6.4 North.~atern States 1960 42.S 48.1 9.d 1970 33. ~ 57 O 6 9O 3 1978 21~5 64ol 1404 = Source s Milton and Hong (1981b) . Sao Jose dos Campos is similar to Sao Paulo, Santa Cruz do Sul to the Southern Region' and Parnaiba to the North- eastern Region. It is also noteworthy that Table 54 shows, both for the country as a whole and for the four regions, a marked increase in free unions between 1960 and 1978. Marriage strategies can also be understood in light of the various types of union in which interviewees found themselves . Consider ing the nine contexts as a whole f irst of all, it can be seen that out of a total of 2, 23 4 unions, the great majority--88 percent--were first unions. Of these, 71 percent were legal unions, while half of the remaining 30 percent were religious and the other half free (Table 551. It ~s evident that, with divorce impossible in Brazil at the time, a new civil marriage was impossible, so that mung second unions, permanent and consensual unions were more frequent. Thus legal unions among second unions refer to cases where Ache

161 a: . - N m ·. al Ed c ~ 0 0 ~ o c o Ed 1 '- MU 0 0 Z e L. ~ P. o . - :D o a a: u ~ 0 Z a: c) 0 0 ::} ~ ~ O a: O C o . - V] `: a 0 al 0 a, ~ ~ 0 · · · · ~ O O tar 0 lo ad ,— Cal CD O to rot ~ O · · O 0 ~ Cal Up Us ~ ~ ~ Cal ~ Cal ~ O go 01 ~ · · ~ lo ~ ~ 0 0 · ~ so 0 fir 0 lo 0 Cal' ~ Cal 0 0 0 ~ ~ O O ~ ~ ~ O In ~ 0 r. 0 ret lo 0 ~— 0 _I ~o ~0 _i ~ ~ ~ ~ ~ a' o m ~ 0 ~ 0 3 O ~ 3 `~' f:: ~ O ~ ~ O O O a: , _t ,~ u =_ ~ ~ c) O : ~:: . - ~ ,~ ~ ~ ~ 0 0 0 o: U ~ U E~ E~

162 f irst legal union ended in widowhood, or the f irst relig- ious or f ree union ended in separation ~ It is iIIunediately noteworthy that at the start of their married lives, 14 percent of women (or of men or of part- ners) preferred no binding connection J and 6 percent did not even choose the bond of living together. This type of preference is a focal point of the discussion in the present chapter. First, the chapter analyzes whether this kind of behavior is more colon in more recent than in older unions ~ Next, it examines whether even the conven tional rural/urban distinction dif ferentiates this beha~r- ior O Finally, these issues will be elaborated by examin - ing each context individually. Table 56 shows the distribution of the total of 1967 first unions of ever-saarried women according to when the unicorn occurred, that is, women rearried by 1960, between 1961 and 1970, and after 1970. As can be. seen, more than half were older unions--before 1960--while 29~1 percent took place between 1961 and 1970; the rest were more recent unions. It is interesting to note that the prefer- ence for legal unions remained at the same level: about 71 percent of first unicorns were legal, irrespective of when they were initiated. Religious unions, however, gradually lost their relative position' falling from 18.4 percent for older unions to 7 0 3 percent for those initi- ated after 1970; there was also a concomitant rise in per- manent and consensual unions, which together accounted for 2lo3 percent of more recent unions. In other words, the points made in the analysis of Tables 53 and 54 are supported by using as a reference point whether the union is recent or not' that is, the different marriage cohorts. PI breakdown of the total sample from the nine contexts into urban and rural shows first that within both groups there is considerable similarity in the temporality of first unions. Indeed' 52 percent and 51 percent calf women were married before 1960 in urban and rural contexts, respectively; for both contexts 29 percent were married between 1961 and 1970. on the other hand, some very interesting differences can be seen in both contexts in r egard to types of unions and thing. As regards the preference for religious union as the first union, in urban contexts, this preference declined drastically over tine, from 14.0 percent to 2.6 percent, a drop of 5.4 times (see Table 57); in the rural areas, ache drop was only 1. 8 times . For more recent rural mart Sages, relig i- ous unions represented about the same proportion (13.8 percent) as that for marriages that took place at least

163 EN Z O O ¢: C O - {R 1 ~4 - o Q4= :>, C ED ~ ' lo 41 1 O 3 lo _ ' to 1 ~ Cal o e 3 F. — o ¢ :' 02 _ N Sit ~ O L. m a <, ·e \0 ~ ~ ~ X An' - m" c A: ~ 0 Ed ~ Cal a, u P. 0 0 Ed Z ~4 P4 m O O ~ · · ~ US ~ Us 0 ~ ~ ~ ~ ~9 Cal ~0 . ~ In cat 1 ° ~ 0 ~ ~ r. U ~ e 0 ~ Pu o z O Cl, a, 0 C U P4 o Z ~r ~ ~ 0 · · · O C~ r" o o a' ~4 o . o o C~ {D ~r ~ ~ ~ ~o · ~ ~ ~ CD ~ ~ ~ o, . ~ ~ 0 ~ ~ ~ C~ 0 o . C~ ~n . ~ ~ ~ £ ~ O O ~ ~ C: _ >, ~ m~ ~ O _' oC O ~ tR ~ ~ ~ `: ." O ~ ~ _, ~ U a~ :, O O ~ C) ~ C Z ~ ~ —~ ~ ~ ~ ~ ~1 ~ ~ ~ ~ ~ C E~ o: C~ & ~ E

164 a) · - sat ~ :r: A: a, o o o ·d I:: ::) e ~ n x ::, A a; O O^ ¢, Cal 1 ED - C: C) O :~ ~ ~ ^ _ O Sat ~4 1 % 1 ~ _ ' O^ to ~ ~0 O :3 ~ · - · - ~ ~ ~ `: N {Q — Ll _ ~ ·. [_ lo ~ ~ m ~ ~ X to 0 ~ m a: i: ~ 0 0 En Ha cat . ¢ al x A: C) m · . ~ d. lo CO ~ 0 · · · ~ an ~ a' us · ~ \0 ~ Cal so O · ~ so ~ Cal Cal c~ ~ o ~ · ~ . o ~ u, ~ c~ a, c~ ~ c~ ~ a' · . ~ ~ \8 (o ~ · — c~ ~ r~ o . ~o o . o . a' . . ~ o o ~ ~ o ~ o o u, ~ · c) . r~ 0 ~ · · · a, O a~ t_ ~ q ~: ~ o ~ ~ ~ o Q . . . . Ll \0 U o - :} a~ o 3 - o ~ ~ ~: o ~ - · - o ~ ~: _ - _ · - U _ C) P~ :' eQ Z `:: U] v ~: :x o ~: t) o o E~ ~

165 15 years ago in urban contexts (14.0 percent). Moreover, in the rural contexts, the preference for legal unions also fell in relative terms, although less drastically. At the same time, there was a moderate rise (and a steeper one than in the urban contexts) in permanent and consen- sual unions as a percentage of all first unions; this reached 26.2 percent of initial preferences for more recent unions. The next step in the analysis was to investigate how these findings operated within each context, that is, to see how the preferences for the various types. of union among the total first unions developed over tome. To this end, Table S8 shows the distribution of ever-married women according to the type of union, for the same three marri- age cohorts. It may be noted immediately that in all con- texts, there was a progressive increase in permanent and consensual unions; only Santa Cruz-Urban showed any sta- bilization in the proportions of permanent unions, while only in Cachoeiro did consensual unions show an increase in proportions, followed by a decrease. In this comparison of contexts, it is highly interest- ing to note that in the country' s poorer areas analyzed here Parnaiba and Conceicao do Araguaia--religious unions represented a noteworthy alternative among unions occurring before 1960; indeed, they competed with legal unions in Parnaiba-Orban and overtook them in Parnaiba- Rural. A number of different reasons may explain this. First' work relations, above all in the countryside, then offered no protection in the form of social security for individuals or their families; thus no documents were required on family structure, birth of children, and so on . Another reason, somewhat related to the f irst , may be the fact that no property was owned that could be divided for inheritance. It is also worth mentioning the possible absence of civil registries at that time in cer- tain areas of the country, whereas churches have always been universally present. Then, too, the continuation of the custom of religious marriage, which is so deep-seated in Northeast Brazil, represented a resistance to the new relative prominence granted the Catholic Church by the Republic. Finally, religious marriage as a first union can be seen as a forerunner of legalized union. The relative decline of religious unions in these con- texts may be due to changes in all or some of these ele- ments. One change is undeniable: access to registries became easier with Improved communications. Moreover, the fact that a legalized union entitled one to social secur-

166 a, 4 ·_. N 4 4 a] o ·. I:: ~ O X ·-o V :D I: o aIJ C) · - c: as O ' rat :^ 1 EN ~ ret O ~ - ~9 ._' o A: . _ o ~ 1 £ - 1 O ~o 3: ~ · - ~ _ x: o 1 ~ . - O `:: - o ~, :J ~o ~ O · - S O ~ L' E~ x: o - N o 4 U] C) o o CD U o - ~ V ~; 3 C) 4! v U) ~ o U m 0 ~e ~ ~ 0 0 C . C~ q, O 4,, N C-} ~ O U' r~ \0 \0 ~4 U' C~ O \0 O 'O o · O U' · · · · c~ ~ ~n e~ ~4 O ~ ~4 · . ~ e~ ~ O ~ 0 ~ r. 01 O \0 `0 ~ ° ~ Ce, t~} ° 0 10 ~ C' · ~ · · 0 0 r~ 0` · · . ~ ~ ~ O C - o ~ ~ C O _ ~ _ _ _ ~ _ ~ ~ C) 3 ~ o ~ o ~ ~ 4` ~ ~ ~ e c C, ~4 D - C 4 - :, o o 4o - C U C~ . o Cr, ~ · . ~ o U~ r~ · · . ~ :s, r~ ~ ~ c~ · ~ · · ~ eo o e ~ ~ ~ C) t: ~ ~Q 4 - C4 P. ~ ~4 U ~ C~ ~ ~ · · ~ — C~ · · · · C, ~ O C~ · · · · C~ O `0 r~ ~o · · . — \0 O O U~ U' · · . ~ O ~ ~n c o - :D - t to ~~i d 0 ~ ~ a' · · · · ·n IO ~ c~ r" · · . ~ t l · ~ ~4 3 -4 ~ ~ `: ~ C o ao :' ~ _ C o _ _ ~ ~ U _ _ ~ ~ ~ ~ P. r~ o rc o :, ~ l: o o U

167 ity certainly represented a stimulus. For example, in Parnaiba, the Social Security Office performs a most important role in the life of wage-earners (Loyola, 1978), many of whom are laid of f for s ickness due to nervous exhaustion. They then reg ister with the Soc ial Secur Sty Office and, as they themselves put it, Stay on the shelf. for months or even years receiving benef its. A new supply of labor is therefore recruited to replace those laid of f . When some of these fresh workers are in turn laid of f s deck, some who have previously been laid of f are reen~- ployed, and so on. Thus, each family group always has the possibility of one of its members being either laid off and receiving a portion of his or her wage from the gov- ernment, or employed and receiving the full wage. Although this situation may seem strange, At represents an attempt to overcome the problem of poverty in the region, or to Redistribute poverty.. At the other extreme low proportions of religious unions are observed in most contexts, except unions initi- ated before 1960 in Recife, where this type of union still represented 11 percent of the total (Table 58). In Cach- oeiro and Sao Jose dos Campos, this proportion was already low, falling to zero for more recent union=. As for legal unions, Cachoeiro practically maintained stable pattern (around 90 percent); Santa Cruz-Orban and -Rural, which accounted for the highest level of legal unions, saw an increase of this type of union from the first to the second cohort, mainly because of a drop in the relative importance of religious unions. In these two contexts, the decline in the relative weight of legal unions among more recent unions can be explained directly by an increase in free unions. It should be remembered that Santa Cruz do Sul is made up Largely of European immigrants, mostly of German origin, who developed family- based agricultural activities and later specialized in tobacco growing. This led the urban part of the munici- pality to develop an industrial complex based on the pro- cessing of tobacco. The prominence of legal unions can be linked not only to cultural factors, but also to issues of property ownership and inheritance. One possible explanation of any decline in the proportion of legal unions may therefore be removal of the latter stimulus by the breaking up of these properties, already small, over the years.- In Sao Jose dos Campo8, where much of the working population is employed on a wage-earning basis by large and medium-sized companies, it could only be expected that legalized unions would account for a high percentage of the total number of unions.

168 The Role of Consensual Unions To understand Brazilian marriage strategies, it is essen- tial to examine the role of Consensual unions more close- ly. Consensual union may be an option for living together when there is no stimulus for legalized union, as discus- sed above. To this reason should be added the far from negligible cost of a civil marriage.8 In Parnaiba, for example, one can clearly observe the phenomenon of ~mar- riage by seduction. (Loyola, 1978) as a strategy to avoid spending money on a civil' or even a religious, marriage. This represents a method used by women--above all those from the poorer groups within the population--to guarantee a union in a context where the male-female ratio in the 15-49 age group was only 84:100 in 1960 and 1970. Once the girl had been ~seduced, ~ her family stoned the ~seducer. and demanded that he remain with the girl To correct his error,. but without a dowry or any wedding feasts or ceremonies, as a pseudo-punishment. Consensual unions are also a solution to separations f allowing legal or even religious unions. Indeed, up until 1978,9 dissolutions of legal unions, even when formalized before a judge (desquite), made new civil unions impossible. The Catholic Church also does not per mit a second religious union, since this bond is indis- soluble according to the law of God. Thus, the only way to marry again was through a free union e Consensual union may also result from a change in values that sees legal bonds as unnecessary. This may be especially important among the younger groups of the pop- ulation' as part of a general questioning of traditional norms and values e Women's liberation has made women both more economically independent and more disposed to express their sexuality, again leading to a preference for f tee unions. The preference for Consensual union may reflect a definitive decision, or it may represent a preliminary stage in conjugal life prior to a legal and/or religious union. It is also probable that pregnancy may contribute to legalization in large Brazilian metropolises. However, the size of the cities included in the NIER, such as Recife and Sao Jose dos Campos, together with the tome covered by the survey (which stopped at 1977), made it impossible to support this conclusion empirically. The issue raised here can be clarified by a brief anal- ysis of Consensual unions for each of the nine contexts. To this end, the following classification has been devel- oped:

169 (1) consensual union as the only union (2) consensual union following a legal union (3) consensual union following a religious union (4) consensual union prior to a legal union (5) consensual union prior to a religious or permanent union The study was made based on the three above-mentioned marriage cohorts, with results as shown in Table S9. It should be noted at the start that the small number of cases for each context, except Recife, permits a merely illustrative analysis of each situation. It must also be remembered that marriage strategies are a dynamic process; thus a woman living in a consensual union at a given moment has a certain probability of legalizing it or of entering a religious union at any subsequent moment. Sao Jose dos Campos is one context where the great majority of consensual unions represented a second or third union following a civil marriage. It thus illus- trates situations (2) and (31. Out of the total of 42 cases, 16 occurred before 1960, a further 16 between 1961 and 1970, and the remaining LO after 1970. For older con- sensual unions, 68.7 percent occurred as the only marriage solution after a legal union. This proportion fell Deco 56 e 3 percent for the following cohort, while for more recent unions it was 60.0 percent. The 6 cases out of 42 in which consensual union preceded a civil union took place before 1971. Of these, 3 cases involved the legal- ization of a consensual union, with 2 cases taking place after the birth of a child; in the other 3 cases, the legal union was with a different person. Situation (1) represented 9 cases, or 21.4 percent of the total; these 9 cases were equally distributed over the three marriage cohorts. Parnaiba-Rural illustrates a certain balance between the types of consensual unions, with the greatest inci- dence for situations (4) and (5} (34.5 percent). AS noted above, religious union has always been highly valued in this region, one of Brazil 's poorest. Thus there were 6 cases in which a consensual union preceded a legal union and 13 in which it preceded a religious union. Of these 6 former cases, 5 were unions legalized with the same per- son, with 3 of these occurring after the birth of chil- dren. Of the 13 cases where religious unions followed consensual unions, 50 percent were with the same person, with Ache bearing of children an associated factor. In Parnaiba-Rural, it is also worth noting that 21 of the 55

170 ..- N m ·. x I:: a u A: QJ a ~ 3 o :) ~ I:: U! w o o o o I: O e ~ c o — ~ ~ W 0 hi a~ C) o 0 ~ ~ 0 0 ~ - V m ad: ~ 0 it_ I: ~~ V 0 0 OR x a C) ~ US ~ UP \0 ~ Cal ~ ~ ~ ~o ~ r~ ~ ~ ~ ~ ~ · e ~ · ~ · · · 0 ~o ~ 0 ~ a, C~ ~ ~ ~ ~ t~ U~ ~ ~ ~ ° ~ ~ ° · · ~ · · · · ~ ~o ~ ~ ~ cr' 0 ~ ~ 0 0' ~ `0 0 ~ O · · · · · · · · ~ ~ 0 ~ u~ O a' ~ 0 C~ ~ ~ ~ ~ U~ - _ O 0w ~ ~ ~ O ~ ~ O O ~ ~1 ~ 1 0 N S N 1 0 ~ ~ ~ O "Q ~2 0 U N (,~ ~ O — — ~ ¢, O O O S:: O ~ ~ ~ U C S~ C ~ It' ~ ~ O ~ ~ ~ ~ (tS u~ ~ 134 C~ ~ tQ cn u~ c~

171 consensual unions began before 1961, meaning that they had already lasted at least 16 years; only 3 of these had occurred following a legal and/or religious union. The picture presented by Parnaiba-Urban is to some extent parallel. Of the consensual unions prior to another union (24.3 percent), 6 cases preceded fecal unions and only 2 preceded religious unions. Of the 6 former cases, 5 were with the same person, with legaliza- tion occurring after the birth of one or more children. Santa Cruz-Rural and Sertaozinho were the contexts that showed the highest proportions of consensual unions as the only union (situation (1)); in both contexts, most of these unions occurred before 1961. Consennual union as a strategy preceding another type of union (situations (4) and (5)) was most infrequent. It was also infrequent as a solution following a separation, perhaps because the separation rates in these two rural contexts were among the lowest for the whole NICER (12.4 percent and 18.S per- cent, respectively for Santa Cruz and Sertaozinho, as com- pared, for example, with 34.4 percent and 39.5 percent, respectively, for Parnaiba-Rural and Recife). In Sertao- zinho, there were 2 cases in which consensual union pre- ceded a legal union, with legalization associated with the birth of a child; in Santa Cruz-Rural, Were were 2 such cases. However small the number of consensual unions in Santa Cruz-Urban and Cachoeiro, where the numerical results were subject to considerable sample fluctuation, it should be noted that these contexts showed a very similar absence of consensual unions as a stage prior to a legal union. This may simply indicate that not enough time had passed for these unions to be converted into legal marriages since, as seen Wove, religious unions were extremely rare in these two urban contexts. However, from the data for Cachoeiro, it can be seen that of the 14 consensual unions as only unions, 7 had already occurred by 1961, while the other 7 had occurred during the past 15 or 16 years. Thus, th:s is not a recent practice. The situation was analogous in Santa Cruz, where for 7 cases, 3 were older and 4 had occurred over the preceding 15 years. The context with the highest number of consensual unions was Recife, which represents 36 percent of the total for these unions. Table 60 shows how the different consensual situations were distributed in Recife for three distinct carriage cohorts. These figures show that the first two marriage cohorts behaved in a very similar way: consensual union as the only union had a greater weight,

172 TABLE 60 Distribution of Consensus Unions, by Type for Three Marr iage Cohorts, Recife: Brazil Marr iage Cohor t Consensual Union Only Type of Union 44.2 42.5 61.1 Before a Legal and/or Religious Union Af ter a Legal and/or Relig ious Union 28.9 27.0 5.6 26.8 30.0 33.3 Total ~ ~ er of Women S2 40 54 from 42 percent to 44 percent, while the remaining 60 percent was equally divided between the other two situa- tions. In other words, almost 30 percent of consensual unions preceded a legal and/or religious union. For the more recent cohort O the situation is the same as regards consensual union after a nonfree union. It can therefore be seen that around 30 percent of consensual unions, irrespective of the marriage cohort, represented a solu- tion to the problem of separation. As for consensual union preceding a nonfree union, this percentage fell con siderably for the last cohort, with a resulting rise in the percentage of consensual unions as only union. This suggests that there had not been enough tome for a con- sensual union initiated between 1971 and 1977 to be trans- formed into a legal and/or religious marriage. Examining the 17 cases of consensual unions preceding a legal union for the three cohorts, it can be seen that in 8 cases, legalization occurred with the same person and following the birth of a child; in 2 cases, legalization was not associated with the birth of a childe Type of Union and Fertil' ty The mean number of children born in consensual unions, while lower in most instances than the means for religious a

173 and legal unions, is still significant in all nine con- texts, as can be clearly seen from the data in Table 61. Except for Cacho-~ro de Itapem~rim and Santa Cruz do Sul- Urban, the mean number of children born per woman was greatest for women in religious only unions, followed by those in legal unions, and, lowest of all, those in con- sensual unions. Unfortunately, the small size of the samples makes it impossible to break down each type of union by its duration. For this reason, although a central point of this report is variation among the contexts, it is useful to combine them in order to increase the sample size and be able to control for the duration of the union, which enhanced the study of the influence on fertility of the type of union. Table 62 shows ache mean namer of children for those women who had had only one union and were still married at the time of the interview. This group was chosen to avoid the problems raised by combining all the children born of different unions that might have occurred within the same period. As can be seen, there is a clear decline in the mean number of children from only religious to permanent unions for the first two marriage cohorts. TABLE 61 Mean Nether of Children Ever Born Alive by Ever-Married Women, by Type of Marital Union, Nine Contexts, lets: Brazil Context - of union Civil and Religious Religious and Only Civil Only Consensual Cachoeiro de Itapemirim 2~2 3el 3~2 Santa Cruz do Sul-Urban 3.5 2.5 305 Sao Jose don Campos 4.2 304 1.2 Sertao Zinho 5.8 4.3 2.6 Santa Cruz do Sul-Rural 5.8 3.9 2.3 Recife 4.4 3.5 2.1 Parnaiba-Urban 5.6 5.3 3.1 Conce$ceo do Araguata- Rural 4.1 3.9 2.3 Parnsiba-Rural 6. ~ 5.3 2.7

174 TABLE 62 Mean Number of Children Ever Born, for Currently Married Women (first marriage), by Type of Union and Three Mar r iage Cohor ts, N ine Contex ts: Br ~ z i 1 r _ Marr iage Cohorts Types of Union Until 1960 Prom 1961-70 A B From 1971 on All C Religious Only 406 (181) 4.4 (128) 1.9 t46) 4.2 Civil and Religious and Civil Only 4~3 (922) 3.2 (708) 1.5 (464) 3.3 Continual Permanent 3.9 (46) 20 ~ (°~7) 1.6 (79) 2. ~ 3.4 (32) 2.? (40) 1.6 (45) 205 Note: Numbers in parentheses are nor of women. However, the difference between the fertility of women who had only religious unions and those who had civil and/or religious unions tended to increase considerably for unions between 1961 and 1970. It should be noted that any conclusion about the most recent marriage cohort must be regarded with caution since this period covers 6 years at the most, and may thus include unions that lasted a very short time; in this period, a difference of, say, one year makes a considerable difference as regards fertility. AGE AT MA=IAGE It is valuable for the study of nuptiality and fertility to estimate some nuptiality parameters, such as initial, mean, and final age at marriage. The information used for this purpose concerns the proportion of nonsingle women and the average par ity per age bracket for 1965, 1970, and 1975. The method used to adjust nuptiality patterns to the empirical data was that suggested by Coale (1971), with r2 defined as the ratio of nonsingle women aged 15-20 to those 20-25, and r3 the ratio of nonsingle women aged 20-25 to those 30-35. Initial age at marriage was conf ined to the 10-14 age bracket . The small number of cases in each context once again made it duff icult to interpret the results, as shown by the data in Table 63. Indeed, these estimates fluctuate considerably from one period to another within a single context, thus restrict-

17S Cat ID O up ~ up ~ ~ or ~ ~ 0 do r— · e · · · e · · e .01 ~ ~ ~ ~ O ~ ~ Cal ~ O _ ~ HI ~ ~ ~ ~ ~ ~ ~ d US ·. - N ~ ~ hi: In lo ~ ~ 0 ~ up a: ~ 0 0 _I O t— ~ 1 1 d' _1 ~ t~ · ~ e e e I I e e cry up ~ ~ ~ ~ ~ [_ ·- ~ ~ ~ U' X . . A ~ \0 _I O ~ u~ t- I- 1 C~ 1 1 1 [- 1 C: ~D ° · 1 · 1 1 1 · 1 ,4 ~ ~ oo a~ 0 C~ ~4 _i ~ ~ ~ ~ ~ 0 ~ ~ a' 0 ~ o, c~ _1 Z ~ t~ ~ ~ e ~ · ~ · ~ ~ co ~ a, . - ~ ' ~ C~ ~ ~ ~ ~ ~ C~ O s: x: ~ ~ a, 0 ~ O C~ ~ C~ _I ~ 1 1 ~ a, IE: ~ ~ . . . . . 1 1 · ~ ~ ~ r~ O cn ~ ~ ~ c~ - ~ ~ ~ ~ a' ~o O S: ~ ~ ~ 1 ~ 1 1 1 ~ ~ J e I ~ I ~ l ~1 ~ ~ ~ ~ ~ ~ C~ . ~ ~ o, O ~ ~ a' :D u~ ~ u~ ~ \0 ~ · ~ · · · . · . U1 ~ ~P ~ ~ ~ C~ O . a, O 0 a: ~ ~D ~ 0 1 1 ~ C~ · . · . · 1 1 , ~ ~ ~ ~ ~ O. O O PO . - ~ ~ ~o ~r ~ ~ ~ ~ 1= ~1 1 o 1 1 1 a~ 1 a' ~ ~ ~ · · ~ 0 1 1 1 · 1 ~ ~ £ ~ ~ ~ c~ z ~4 o ~q .,. x t) · _1 ~, o o o N a~ ~ ~ ~ 0 1 1 3 Q~ _I —I ,q U) C: ~ ~: O O ~V ~ m ~ ~ ~ ~ 0 0 0 ~ ~ 0 ~ ~ N :' ~ ~ S 0 0 ~ `4 ~ ~ ~ ~ ~v U! C} O .O ~ N t O —~ ~ O — O ~ ~ C ~ ~ U aC O ~ ~ ~ ~ ~ C ~v to ~ tU ~ 0 0 cn u~ u~ P~ ~ u~ C) c~ c V - t e. o

176 ing any analysis of trends. This effect of fluctuations in the Samples is reduced by working with data for al 1 the women in the household rather than just one woman per household, as has been the case until now in this discus- sion. In fact r this leads to a substantial increase in the size of the sampl es for each context; unfortunately, however, it is only feasible for the year of the investi- gation, for which information on all members of the house- hold is available. Table 64 shows mean age at marrisqe calculated by Hajoal I s method (1953), on the basis of information about single women in the households by age bracket. As can be seen, the f indings for 1975-77 are generally very similar to those for the various regions of Brazil (see Table 7) in 1976; this suggests that the quality of the data is acceptable and that the fluctua- t ions observed in Table S7 were in fact due to the small number of cases. Moreover', the effect of migration must also be present as a disturbing element in this type of analysis. All the contexts were indeed subject to migra- tory influxes of varying intensity and durations affecting the contingents of men and women apt to marry.1 With these reservations in mind, it can be seen that the highest initial age at marr iage in 1975 was found in Sao Jose dos Campos (14.9 years) and the- lowest in Sertao- TABr~ 64 Mean Age at Marriage (Hajnal method), Nine Contexts : Brazil . Context Mean Age at Marriage Sao Jose dos Campos Santa Crllz do Sul-Urban Santa Cruz do Sul-Rural Parnaiba-t~rban Parnaiba-Rural Sertaozinho Conceicso do Araguaia Cachoeiro de Itapemirim Recife 23.74 2S.22 23.37 23.39 23.47 23.42 19.64 23.9S 23.26

177 z intro (10.2 years). These extremes seem consistent with the mean for Brazil in 1976, estimated by Altmann and Wong (1981b), at 13.3 years. For the Northeastern region, the age found was 12.7 years, very close to the values for Recife (12.8) and Parnaiba-Rural (12.7). For 1970-75, even taking all the above reservations into account , i ~ can be seen that, in general, there was a certain increase in the initial age at marriage. As regards the f Anal Able at marriage r an important index of nuptiality and closely related to the Marriage market,. the larger urban centers and metropolises were the regions that most favored late marriage; this applies to Recife, Sao Jose dos Campos, and Santa Cruz do Sul-Urban, contexts in which final ages were highest in 1975. With some exceptions, it can also be said that the final age at marriage generally increased between 1970 and 1976. It should be observed, on the other hand, that the contingent of women who remain single after a given age depends, among other things, on the availability of men Exposed to the risk. of forming a union. In 1970, the sex ratio varied considerably in the 15-49 age bracket for all the areas studied. If the information at the level of municipalities from the 1970 census is used, the sex ratios per 1, 000 women aged 15-49 were as follows: Ser- taozinho, 1,115; Sao Jose, 1,040; Santa Cruz, 970; Cacho- eiro, 960; Parnaiba, 953; Recife 0 785. The proportions of women still single in the 30-39 age bracket, as shown by the data at the household level, were, in the same order, as follows: 7.4 percent, 8.8 percent, 9,1 percent, 12.3 percent, 12.5 percent, and 14.8 percent. In other words, the higher the sex ratio, the lower the proportion of still single women. Given the reservations mentioned above, it may be con- cluded from this descriptive analytic that, despite the different levels in the parameters for the various con- texts, initial age at marriage is rising, mean age at marriage is also increasing, and final-age at marriage has also risen. What is far less clear from these data is the timing of the changes, which makes it difficult to draw firm conclusions about how they affect fertility levels. In any case, as indicated in the following chapters, the primary factor involved in Brazil's accelerated fertility decline is declining marital fertility, traceable to changing patterns of contraceptive use.

CHAPTER 8 MARITAL FERTILITY The total fertility rates analyzed up to this point reflect the effect of unmarried women' that is, they depend on a larger or smaller proportion of still- unmarried women En the various age brackets. In the 30-39 age bracket, the figure may be as high as 15 percent. Moreover, as noted above, this proportion varies from one context to another. Marital fertility as measured by the total marital fertility rate, the topic of the present chapter, by definition avoids this effect. Table 65 shows total Marital fertility rates for 1970 and 197S. In 1970' they varied from 3.70 to 9.61, and in 1975 from 2.96 to 9.64. In both cases' the lowest figure is for Santa Cruz do Sul-Urban, and the highest for Parnaiba-Urban and -Rural. As can be seen, except for Parnaiba-Rural, mari- tal fertility declined in all contexts between 1970 and 1975. The largest decline was registered for Sertaozinho {as noted above in the comparison of total fertility rates). In those populations where birth control is not practiced, the natural marital fertility pattern after 25 years of age is shown by a convex curve that drops slowly until age 35, and then falls abruptly to reflect the steep decrease in the proportion of fertile women. On the other hand, in populations that voluntarily control fertility, the rapid decline in marital fertility rates at early ages results in greatly reduced levels of fertility at the age of 3C), giving the curve a concave shape. Thus, the decline in world fertility levels involves a transition from a structural pattern represented by a convex curve to a new pattern represented by a concave curve. In Bra- zil, this transition can be seen in both high-fertili~y and low-fertility regions. This transition can be seen from Figure 6 (see page 40) by comparing the marital fertility curves representing 178

179 TABLE 65 Total Marital Fertility Rates, Nine Contexts, 1970 and 1975: Brazil Total Marital Fertility Rate Percent Decline 197 5 Context 1970 1975 1970% Parnaiba-Rural 9.41 9.64 ~ 2. 44 (228) (243) Sao Jose dos Campos 6.35 S.26 17.25 ~ 417) ( 520 ) Recife 7.26 6037 12031 (386) (472) Conceicso do Araguaia 8.17 7 0 09 13 0 26 (231) (292) Parnaiba-Urban 9. 61 7 o 46 22.45 (231) (245) Sertaozinho 6. 69 4.72 29 c 42 1223) . (248) Cachoeiro 5.32 4 o 16 21.76 (197) ( 250 ) Santa Cruz-Rural 6.68 5.98 10.46 (225) (245) Santa Cruz-~rban 3. 70 2.96 20.02 (218) (234) Note: Numbers in parentheses are number of women. the six contexts with highest and lowest fertility levels in the NIaR. The curve for Brazil as a whole represents the middle point in the transition from a high level as in the case of Parnaiba-Rural (with a roughly convex curve)--to a low level--as in the case of Cachoeiro de Itapemirim (with a distinctly concave curve). These differing marital fertility patterns reflect varying distances from a natural-fertility pattern. Such distances occur in a typical manner and, according to Coale, can be measured through the parameter ~m,. known as the degree of fertility control, or how far fertility practice has moved away from natural fertility as a result of the use of contraceptives and abortion. The degrees of control estimated for the nine regions in 1970 and 1975 are shown in Table 66. As can be seen, in both 1970 and 197S, the greatest degrees of control were in Santa Cruz do Sul-Urban and Cachoeiro de Itape- mir~; in 1975, the value for the latter was higher than for the former. The lowest levels of control were in Parnaiba-Urban and -Rural and Conceicso do Araguaia. The evolution over time of these parameters shows that there was a clear increase in fertility control for all regions.

180 TABLE 66 values for the Fertility Control Measure (m) Estimated by Coale's Method, 1970 and 1975: Brazil Context 1970 1975 Sao Jose dos Campos 0O7495 1.1792 Santa Cruz do Sul-Urban 1. 9284 lo 3880 Santa Cruz do Sul-Rural 0-0451 0O6624 Parnaiba-Urban -0.2444 0.2220 Parnaiba-Rural 0.0486 0.2361 Sertaozinho Conceicso do Araguaia Cachoeiro de Itapemirim Recife -- 1.2004 -- 0.1487 0.6604 1.4S47 0.0520 0.6721 Note: -- indicates sample size too small to calculate. The exception was Santa Cruz do Sul-Urban, which already had fairly high levels, where the degree of control fell from 1.9 to 1.4. This decrease was probably ache main factor leading to that region's increase in the total fertility rate from 2. S to 2.9 children between 1970 and 19750 The variations in fertility levels between 197Q and 1975 can be clarified by comparing them with the nuptial- ity parameters and degrees of control for the same period. In the case of Sao Jose, for example, the drop in fertil- ity was from 4.8 to 4.0 children between 1970 and 197S; this is associated with an increase in the degree of con- trol from 0.7 to 1.2, and with an increase in mean age at marriage and initial age at marriage from 21.2 to 23.S and lle83 to 14.90 years of age, respectively.

CHAPTER 9 THE PROXIMATE vARIABL2S The factors that directly influence fertility and together determine its level--the so-called proximate variables-- can be grouped as follows, according to Bongsarts (1983): al b) c exposure to regular sexual relations - proportion of married women prevalence of deliberate control of mar ital fertility use and efficiency of contraceptives prevalence of induced abortion ) determinants of natural marital fertility duration of postpartum infecundity f ecundity spontaneous intrauterine mortality prevalence of permanent sterility Ilowever, not all these f actors have the same impact ire determining discrepancies between natural fertility and the total fertility rates observed In a given context. In fact, the proportion of married women, the use and efficiency of contraceptives, the prevalence of induced abortion, and the duration of postpartum infecundity account in general for over 95 percent of those discrep- ancies; the effects of the remaining three factors are difficult to measure in such studies as fertility surveys. Postpartum infecundity depends in turn on the length of sexual abstention following delivery and on the duration of breastfeeding. In Latin America, prolonged abstinence Is not as common a practice as it is in some African and "fan countries. As can be seen in Table 67, in the NIER, a limited number of women declared they abstained. On the other hand, what does still exist among some groups of women of rural origin is Postpartum quarantine. or ~re- 181.

l 182 - 3 o U] a) Did E~ Q o . - C) V ~: o U - m ~: s~ s" C) 1 s9 O ~ 3 N S~ ~: ~o a~ X o ~: S~ OQ 5: ~C ~: 1 - 1 ul · 1 · . o, ~ q. o O ~ O ~D o ~ o ~ .-, _ ~ 1 ~ ~ ~ ~ tD :— s~ O 5, ~ · 1 . . ~ . . . ~ . . O ~ ~ ~ ° o ~ ~ r~ o o o un — Cl, u~ ~ O ~r C) o _ ~J ~ c~ ~ r— ~ 1 ~ O u~ ~ ~ O O :1 ~ 1 · · 1 ~ ~: ~ ~ u' ~ ~0 ~ tD ~ ~4 0 c, ~ ~ ~r O ~d ~ ~ _ U] N - _' C P. ~ ~ C C ~ ~ ~2 P. ~4 :, C~ C L. N :' Ld C) ~S 0 C #e CD O ~ Q s~ O ~ O CO ~ C} g' ~C \0 1 0 ~ ~ 1 r' {~ ~ 1 0 · 1 · ~ ~ 1 ~ · ~ 1 o ~ o ~ ~ o o A —I 1 ' ~ ~ 1 1 ~ ~ ~ 1 · 1 · 1 1 1 · · · 1 a, U~ O \0 1 un co a~ 1 ~ ~ 4~ Y~ O · 1 · · · 1 c · . · 0 U~ u~ 0 O U' _l U 1 1 ~ C~ c-4 1 ~ ~ ~ 1 0 1 1 · · · 1 · · · 1 O ~ `0 ~ O C~ ~ ~ O ~ O . I 1 1 1 1 o . O O a' 1 ~ ~ ~ ~ a, ~ ~r ~D O · 1 · · . . . . .. . .. ~ O ~ ~ ~ O \0 0 r~ ~ r4 :s -4 ~ . ~ 0 40 _ C) Ea ~ ~ e C ~ o ~ ~ =, ~ . N ~ O `4 ~ — 4 C) — ~U h4 — e o. ~ ~ ~ _ c a ~ ~ ~ ~c 2 a ~ 8 c, ~ P. cn ~ 0 ~ z

183 elusion, ~ referring to a period when diet, bodily hygiene, and sexual practices are carefully controlled. Given these complications, in this section the duration of post- partum infecundity is measured exclusives y by the duration of breastfeeding . In the terms of the Bongaar ts index, `]escr ibed in Chap- ter 1, the values for the mar r iage rate, Cm in each of the nine NIHR contexts were the following: Cachoeiro de I tapemir im Santa Cruz-Urban Sao Jose dos Campos Sertaoz intro Santa Cruz-Rural Recife Parnaiba-Urban Conceicso do Araguaia Parnsiba-Rural 0.648 0.710 00657 0.842 0.698 00651 0.766 0.777 0.832 The subsections below discuss the NIHR in relation to the other three primary proximate determinants: noncontracep- tion use (Cc), postpartum infecundability (Ci), and abortion (Ca). CON=A=:PTI VE USE Contraceptive Prevalence The NIER registered greatly varying levels of contracep- tive use in the various contexts. Parnaiba-orban and -Rural--an area which, as noted in the Appendix, has been occupied for a long time and now has a practically stag- nant economy--was shown to have the lowest fertility rates consistently, and it ~s this area which reqistered the lowest proportion of contraceptive use. are far from negl~gible, however: ~.~_ _ ~ ~ · ~ _ ~ _ . ~, These proportions indeed, of every five w~men ~n rarna1Da-Rural WhO were married at the time of the survey, one was using some contraceptive method. In Conce~cso do Araguaia, this proportion reached a~most 30 percent, wh~le in the other contexts it was over 50 per- cent. If these proportions are compared with those obtained by the Contraceptive Prevalence Surveys (CPS), described in Chapter 1 of this report, there is a high degree of consistency among some of the available results. For the state of Pisui, except for the capital, the CPS showed a contraceptive use rate for 1979 of around 28.8

L84 percent, very close to the value obtained by the NIHR for Parnaiba-urban ( 27. 7 percent), as shown in column 1 of Table 68. The C]?S rate for Recife in 1980 was 51.5 per- cent using contraceptives, while the NIHR rate was 57. 5 percent. Similarly, in Santa Cruz do Sul-Urban, the CPS showed 63. 0 percent of marr fed women using contraception; this value was very close to that found by Bathes (1976), in 1975 for Rio Grande do Sul (64. 6 percent) . On the other hand r attention should be paid to the fiery high rates found in the NIBR for Sao Jose dos Campos ( 74.9 percent) and Sertaozinho (69~2 percent) D when compared with the data for Sao Paulo found by Nalcamura et al. ('979), in 1978 (63~4 percent for the municipality of Sao Paulo, and 58. 6 percent for the rest of the state) . Even when the NI}IR rates were estimated as an average of age- specific use rates (column 2, Table 68), Sao Jose still had a high rate of 71.9 percent, and the rate for Sertao- z intro remained unaltered. Since there is as yet no reason to suppose that these two contexts constitute special cases in the use of contraceptives as compared with other parts of Sao Paulo state, it is probable that the NIER has overestimated the proportions of contraceptive use. For the purposes of all the -following cat adulations, therefore, the proport ions of contraceptive use for Sao Jose ~ . .d Ser - taozi~o adopted were, respectively, 6304 percent and 5806 percents TABLE 68 Currently Married Women Aged 15-49 Currently Using Contraception, Nine Contexts: Brazil Average of Currently Age-Specific Number of Context Using (percent) Use Rates Cases Sao Jose dos Ca~os 74~9 71.9 415 Santa Cruz-Orban 6300 60.6 219 Santa Cruz-Bural 58.3 54.3 242 Parnaiba-lJrban 27. 7 27.7 220 Parnaiba-Rural 20. 7 19.6 . 207 Sertaozinho 6902 69.3 224 Conceicso do Araguaia 29.6 29.1 214 Cachoeiro do Itapeoirm 69.9 64.5 229 Recife 57.5 52.0 348

185 The data used as a basis for calculating average age-specific use rates showed that control was differential according to age but existed for all ages. Evidently, when observing this fact by age bracket, caution should be the rule given the extremely small number of women in each bracket. To provide a more comp] ete idea of what may be happening in the urban areas of Brazil, the information on all f ive urban contexts has been combined by age bracket. As was done in Chapter 8, this is a device to increase the size of the samples, for as already stressed at the outset, it is contrary to the very objective of the NIBR to aggregate situations as distinct as these. The relevant f igures are thus the following: Age 15-19 20-24 25-29 30-34 35-39 40-44 45-49 Mean value Percent Use 48~1 6200 71.6 71.6 64.3 53.1 26.8 56.8 In other words, 48 percent of young couples living in urban areas are already regulating fertility. The m^Yi- mum use rate is in the 25-29 and 30-34 age brackets, with the proportion falling thereafter. Proceeding the same way for the rural contexts, a lower overall mean value of 45.5 percent is obtained. For the different age brackets, the rural use rates are as follows: Age 15-19 20-24 25-29 30~34 35~39 4 0~44 4S-49 Mean value Percent Use . 30e 7 5 0 e 4 S5e 8 5 3 e 4 46e 9 44 e 0 3 7 e 5 5 e 5

186 Contraceptive Method s Table 6 7 shows the methods used by those women who were resorting to contraception. Such devices as the IUD and d iaphragm were practically absent f rom all nine contexts. There were four main methods: rhythm, coitus interruptus, the pill, and female sterilization. In Santa Cruz-Urban Cachoeiro, and Sao Jose dos Campos, 85. 4 percent, 84. 8 percent and 78. 1 percent of women, respectively, used highly eff icient methods ( the pill or ster ilization, with the highest proportion for the pill)o In Corlce~cao do Araguaia and Parnaiba-Rural ~ 84. 5 percent and 85. 4 percent of women, respectively f were using the pill or tubal sterilization (with a higher proportion for the latter). The remaining f ire contexts showed proportions for use or these two methods ranging from 61 percent to 69 percent; in Recife, female sterilization tonic the place of the pill. In sum, the more traditional methods have already lost ground to more modern and eff icient methods. One interesting fact is the 8.3 percent f igure for use of the condom in Sao Jose dos Campos, whereas this method has practically disappeared in the other contexts; this may reflect the presence of large numbers of Emigrants who had arrived in Sao Jose over the years immediately pre- ceding the survey. Similarly, the high proportion of women using the rhythm method in Santa Cruz-Rural may be explained by the fact that most of the population is of German extraction, and thus inf luenced by the European tradition. The fact that vaginal methods only appeared significantly in Recife is another noteworthy point. Although the NIHR samples were small and the propor- tions of women sterilized at the time of the survey were therefore sub ject to considerable sample f luctuations, the growth of this practice can also clearly be seen in the results of the CPS, as reported in Chapter 1. One note- worthy fact about "he age distribution of married women who were sterilized at the the of the NIER (Table 69) is that in Cachoeiro, Conceicao do Araguaia, and Parnaiba- Rural, 50 percent or more were age 34 or younger, while in Araguaia, L5 percent had not yet reached the age of 2S. What is surprising in all this is the fact that female sterilization is illegal in Brazil. The Medical Code of Ethics, Chapter HI, Con Medical Responsibility, article 52, states:

187 ·e U] X C o C) it: ·~' it a' 3 N I' U) ~: S~ o o 53 eq ~0 - ; U] o U' C o C) o ~ o U2 - fo P4 o: u q a' C ~ :1 U~ a: o o t) t:: o U o o o s C) - u ~: n - ~ c d: 0 h4 ,~ 48 4 o. a L. C) 4 ~ `: D U! ~ :, o C, 5 0 0 r~ r~ O ~ D ~ ~ _ O ~ ~ ~ \0 · . · · · · ~ O U~ ~ \0 ~ ~ U' ~1 ~4 ~ — 0 ~ ~ r~ ~ ~ ~ ' a 0 ~ a' ~ O1 ~— <" ~ ~ ~ r~ ~ c~4 O C~ ~ ~ ~ O4 0 ~ r~ ~ ~o ~ r~ r~ O ~ ~ O O U' ~ 0 c~ c~ 0 ~n t~ r~ 0 O ~ -4 C~ ~ `0 ~ · ~ · . · . ~ 0 ~ ~ ~ ~ ~ C~ o, ~ - 4 ~ ~ \0 0 0 cr' ~ ~ ~ ~ · . · . ~ · ~ O 0 ~ 01 ~ ~ ~ - 0 .0 C - o O O ~ ~ · · · · · 4, O 0 O 0 0 I_ · . . . . . . O 0 0 \0 ~F 01 ~ 01 ~ ~ ~ ~7 ~ O ~ O C~ C~ o U~ ~ Z

188 Sterilization Is condemned, but may be practiced in exceptional cases, when there is a precise indica- tion approved by two medical doctors consulted In conference. Aside from the few cases in which a woman's health prob- lems lead to the advisability of sterilization, the marked increase in tubal sterilization among Brazil's female population may be the result of two factors. The f first is a consequence of the misuse of caesareans over recent years, this has been stimulated by the INAMPS (National Health and Welfare Service) through higher remuneration for this kind of operation than for normal delivery. Official data show that from 1971 to 1980 in Brazil, the proportion of caesareans to total deliveries rose from 14.6 percent to 2903 percent. The rates for Sao Paulo over the same period rose from 17.5 percent to 36.0 percent, and similar increases occurred in practically all the Brazilian state. A study performed in Ribeirao Preto, Sao Paulo state, concerning standards of assistance for deliveries showed that in 1972, 1973, and 1974, 18.1 percent, 23.0 percent, and 24.1 percent of hospital deliveries' respectively, were caesareans (Boletin de la Oficina Sanitaria Panamerican, 1978). Recent data on Greater Sao Paulo and Greater Recife permit a comparison between the type of delivery of the latest and penultimate children born to mothers who gave birth over the last 8 months (Berquo, 1981). The results for Greater Recife show that 14 percent of deliveries of the penultimate child were caesareans, while the rate is 27 percent for the latest childo I~ addition, 11 percent of these mothers gave birth by caesarean in both cases. For greater Sao Paulo, 21 percent of deliveries were caesar- eans for the penultimate child and 31 percent for the latest, while 17 percent of these mothers had given birth by caesarean in both cases. It is well known that any woman who has already had two or three caesareans is a sure candidate for sterilization. The policy adopted by INAMPS has thus increased women's chances of being ster- ilized. Nowadays, in fact, doctors receive the same remuneration whether a delivery is normal or caesarean; nevertheless, it seems that the doctors themselves have grown accustomed to caesarean deliveries and thus continue persuading pregnant women to accept them, with a continu- ing effect on sterilization levels. The second factor which may determine the marked increase in tubal sterilization is linked to the non-

189 off icial family planning programs underway in Brazil. As noted in Chapter 3, physicians in private practive have been mak ing more f requent use of public f ac ilities to per- form sterilizations even though use is not officially promoted. If this were not so, given the legislation mentioned earlier, these results would not be occurring and affecting even. very young women, as seen in the data shown here. Bongsarts' Index of Contraception In calculating the Bongaarts contraception index (Cc = 1 a S x E x U), described in Chapter 1, the values for "U. are already available in the second Cole of Table 68. Values for ~E. in any given context are calculated by adding the products of the values in Table 63 and the weights which correspond to each of the methods in accordance with its Off iciency: Methods Foam/Cream Jelly D iaphragm Rhythm Condom IUD pill Sterilization Abstention Other Contraceptive Ef f iciency 0~87 0~88 0 a82 Oe91 0 a96 0~98 1 e 0 0 1~00 O.90 Values for "E. thus calculated are shown in Table 70. If BUT, ~E., and "S. are replaced in the formula for Cc by their respective values, the results obtained are as shown in Table 71. As can be seen, the degree of control is greatest in Cachoeiro de Itapemir~, followed by Sao Jose and Santa Cruz do Sul-tirban. The lowest values of Cc are for Conceicso do Araguaia and Parnaiba-Urban and -Rural. In other words, these results conf irm the f indings Obtained when Imp was calculated by Coale's method. The relation between the values for Cc and total fertility is also very clear : that is , the lower the value for C<, the lower the total fertility rate.

190 `;e S" 3 o - ~n a, U S~ o o C - U e~ W 0 :D o V tQ ~ JJ Q`t X C - o ~ C'-) m ~S 2 ~: a~ Q~ 3 o r ~, ·- ~; ~ E~ ~ N L. Z Ct 1 - a' z c ~ O cn N a 1 c ~ 0 U1 C. 0 C, 0 oo ~ 0 U) ~ 4, - u ~4 - 0 ~ Ld 4, _ 9, 0 1 r-1 ~ O C) ~ O C O O ~ O O ~ D t" "e iS, O _. O O O '= ~ · . e 0 0 ~ ~r _ ~ ~ ;~ O O O O ~ O O ~ O 0 ~e ~1 0 1~ 0 Y. _' O · e ~ e ~ O 0 ~1 0 —~ O ~1 ~ O \0 _. ~ C~ O CD re f., e e e C~ ~ ~ ~ O O O O 0` u~ 0 ~o a 0 e · e · ~ O ~ ~ ~ t_ 01 O O ~ U~ ~ O u~ Od S~ tD O ~ ~ O O · · · 0 · · · · · ~ O ~ `0 ~ O O ~ \0 ~ `0 U' - 0 60 41 4|| O O ~ee ~ O O U~ ~ ~ ~ ~ O1 <" O o `0 t. N O r~ ~ · e o o o e e e ~ ~ ~ tO ~ _' O O ~ ~ ~ ~ C - N o ~4 0 CO - a -4 C Ll L ~: ~ C C h4 J o4 a O 4, x~ o' ~ ~ ~o r~ en \0 · . · — ~4 ~ ~ U' -4 ~ 0 0 -4 0 0 0 ~ ~ 0 U' ~ o, ~ 0 0 0 U' ~ 0 · . · . · . · . · — {~ ~ C. O O O C~ "e O re 01 O t~ O ~ O ~ ~ tme O ~ O ~' ~ ~ {q · · · · · · · · - O ~ O C ~ O C,) c ~ ~ 4~1 O ~ ~ ~ ~ — ?4 ~ ~ O _ C 0 — O ~C ~ C ~d 4 L4 ~ ~ O. t0 ~ C ~ ~ > L. ~ O _ O ~ ~n ~ ~ u a ~ 7e 0 >.

191 TABLE 71 Index of Contraception, Nine Contexts: Brazil Context Value of Cc Cachoeiro do Itapemirim Sao Jose dos Campos Santa Cruz-Urban Sertaozinho Santa Cruz-Rural Recife Parnaiba-Urban .312 .316 .360 .381 .450 .457 .710 Conceicso do Araguaia .693 Parnaiba-Rural .791 Postpartum Infecundability Before the 1960s, almost no studies on breastfeeding in Brazil were carried out. Rea (1981) presents a table, reproduced here (Table 72), which gives a good idea of the Brazilian breastfeeding situation. These data show considerable variability depending on the region and liv- ing standard. The author does, however, emphasize that comparison in this case should be approached with caution since the information and/or methodologies involved are mostly heterogeneous. As noted in Chapter 5, the phenomenon of early wean- ing exists in today's Brazil, as it also does in a number of other Latin American countries (Lesthaeghe et al. 1981). On the local level, a recent study conducted by UNICEF/INAM/~-~:uRAP (Berquo et al. 1981) in the metropol- itan areas of Sao Paulo and Recife showed a very low aver- age time for breastfeeding for mothers in both areas. Unbiased estimates of the aver age duration of natural breastfeeding can be made using the calculation technique known as the current status method, based on the propor- tions of children in each age group who were being breast- fed at the time of the interview (Jain and Bongsarts, 1981). The average time for Greater Sao Paula was 3

192 TABLE 72 Percent of Breastfed Children at 4 Months of Age, by Various Studies: Brazil Location Year Percent Remarics Recife (PE) 1968 5.8 >3 months Sao Paulo (SP) 1971 72.0 immigrants Sao Paulo (SP) 1973 31.6 low-income Ribeirao Preto (SP) 1974 39.3 Salvador (BA) 1974 38. 9 Rio de Janeiro (RJ) 1974 25.0 Pelotas (RS) 1974 160 ~ Icapara (SP) 1975 78 .8 3 months, rural Vale do Ribeira 1975 150 0 3 months, rural Salvador (BA) 1975 45. ~ >3 months Getulis" and Guaiembe (SP) 1975 89. ~ Japanese extraction, 3 months Campinas (SP) 1976 35. 3 Austin (RJ) 1977 54.0 Paullnea (SP1 1977 40. 0 Sorocat~a (SP) 1978 30.0 >3 months Botucatu (SP) 1978/79 50 0 7 3 months, medical post, school Sao Paulo {SP) 1979 41.9 low~incame Sao Mateus (SP) 1980 13.0 working~class district Greater Sao Paulo (SP) 1981 38.0 Greater Recife (PE) 1981 2S. O Source: Bea (1981). months, and for Greater Recife 2.2 months. Even conside~- ing that this study included mothers of children no more than 8 months old at the time of the interview and thus truncated the distribution with regard to higher breast- f ceding times, these values are very low. Between the first and second month, 42 percent of children in Recife and 53 percent in Sao Paulo had already been weaned. The fact that women in Recife breastfeed for less t ime than those in Sao Paulo had already been highlighted by Puffer and Serrano tl973) at The time of the Inter- american Study on Childhood Mortality, conducted between 1968 and 1970. This survey showed that of the children who died in their f irst year, the proportion breastfed for

193 one month was 26.8 percent and 35.5 percent for the muni cipalities of Recife and Sao Paulo, respectively; the proportion of breastfed children who died at ages 6-11 months or over was 1. 4 percent and 4.1 percent for R-cif e and Sao Paulo, respectively. Although the data presented by Puffer and Serrano refer to children who had already d fed at the time of the study and thus to some extent underestimated the breastfeeding times, they nevertheless indicate the regional differences involved. In 1975, working with 593 families in the Vale do Ribeira (coastal region of Sao Paulo state) having a total of 1,005 children under 5 years, Rea (1981) found 14.6, 7 . 4 , and 5. 8 months to be the mean breastfeeding times, respectively, for mothers born in the Vale region, other regions of Sao Paulo state, or other states . For the NIT contexts, the mean breastfeeding times were as follows :1 Cachoeiro de Itapemirim Conceicao do Araguaia Sertaozinho Parnaiba-Rural Sao Jose dos Campus Santa Cruz-Rural Parnaiba-Urban Santa Cruz-Urban Recife - 8.5 months 805 7.3 months 6.3 months 6.0 months S.5 months 4.8 months 402 months 3.3 months months On the basis of these values, Bongearts' index of post- partum infecundability (Ci) was calculated. In the NIER survey, as usually occurs in surveys of this kind, there are no reliable data on the postpartum amenorrhea period, that is, on ~i.. For this reason, the model put forward by Bongearts (1983) was used, which establishes an expo- nential relation between ~i. and the mean duration of breastfeeding, here represented by B: i ~ 1.753 exp. ( .1396B - O. 001872B21 . If B is replaced by the values found for mean breastfeed- ing times, the values shown in Table 73 are obtained for .~. and Ci. Since Ci is 1 in the complete absence of lactation or of postpartum abstinence and tends toward zero as the duration of postpartum infecundity increases, it can be seen that Cachoeiro and Conceicao do Araguaia showed the highest intervals for postpartum amenorrhea, while Recife showed the lowest.

194 TABLE 73 Values of Air and Ci, Nine Contexts: Brazil Value of n i. Context in `Months Value of C Cachoeiro do Itapemirim 5.0 0.850 Conceicao do Araguaia 5~0 0.850 5ertaozinho 4 . 4 0. 87 3 Parnaiba-Rural 3 ~ 9 O. 892 Sao Jose dos Campos 3.8 0.897 Santa Cruz-Rural 3e6 0~906 Parnaiba-Urban 3 O 3 0 . 918 Santa Cruz-Orban 3OO 0.928 Recife 2.7 0.942 ABORTION The information on abortion in the No" covers both span taneous and deliberate abortion, for the same reasons as those alleged in the majority of surveys on human repro- duction: the questionable credibility of replies to ques- t ions intended to identify the two separately when the survey itself kiss not been specifically designed to study abortion. With the information available ~ total abortion rates (TAR) were calculated (Table 74). These rates were extremely variable from one context to another. The rate for Santa Cruz-Urban was only 9.2 percent of that for Parnaiba—oral. In Santa Cruz-Ur},an and -oral, abortion was practiced little, with Cachoeiro next in frequency. Sao Jose, Sertaozinho, and Recife had much higher and very similar rates. Parnaiba had the highest rate of all, though lower for urban than for rural women. To calculate Ca, the total fertility rates were taken from Table 51 and values for But from Table 62. values For Ca (Table 74) were all very high, and close to 1. Thus, although this practice is underway in Brazil and involves many thousands of women, in overall calculations of fertility control it is of secondary importance.

195 TABLE 74 Total Abortion Rates (TAR) and Bongaarts' Abortion Rate (Ca), Nine Contexts : Brazil Context TAR Ca Cachoeiro do Itapemirim 0.719 0.963 Santa Cruz-Urban GO 068 0.985 Sao Jose dos Campos 0.462 0.931 Sertaozinho 0 ~ 454 0.942 Santa Cru2-Rural 0.104 0. 987 Recife O O 478 0.947 Parnaiba-Urban 0. 617 0. 9S3 Conceicso do Araguaia 0.263 0.983 Parnaiba-Rural 0.735 0.967

CHAPTER 1 0 SoCIOECONoMIc FAC robs: FAMILY INCOME As pointed out in Chapter S. the fertility rates observed for each of the nine NIER contexts varied considerably. Succeeding chapters showed that, at least at this level of analysis, a variety of reproduction strategies have been adopted by the populations of the different contexts. The present chapter incorporates into the analysis the socioeconomic variables. Among these, family income appears to have distinctive importance and is the focus of the discussion that follows. The analysis below is limited to the urban contexts s ince it addresses only monetary 7 Rime _^ ~ , ~ ..._, Families have been classified in four categories, depending on their per capita monthly income at the time of the survey (between 197S and 1977): (1) up to one-half the minimum wagel3 (2) between one-half and one minimum wage (3) between one and two times the minimum wage (4) over twice the minimum wage This var table has been incorporated into the analys i s according to its equivalent at the particular moment of the time of interview; this precaution was taken since th information concerned was not taken from the individual life histories, but frown data included is the form dealing with the domestic group. Although this disparity makes interpretation of the results somewhat problematic, the present analysis is only preliminary; later research will be able to incorporate variables from the life histories. The distribution of families according to income (Table 75) varies considerably from one context to the next. On the one hand, Santa Cruz ~ Cachoeiro, and Sao Jose dos Campos are somewhat similar, showing a degree of equil 196 . . _

lg7 TABLE 75 Percent Distr ibution of Women Aged 15 Years and Over, by Per Capita Monthly Income ( in fractions of one minimum wage), Five Urban Contexts, at Time of Survey: Br az i 1 Context Up to Between 1/2 Between One Over 1/2 and One and Two Times Twice Hin. Wage Min. Wage Min. Wage Min. Wage Between 1/2 Between One and One Santa Cruz-Urt~an 2100 30~9 24~0 24~0 Cachoeiro do Itapeeirm 22.9 31~0 24.4 21.7 San Jose doe Campos 25;~1 29~3 21~6 24oG "cite 42.9 23.3 1704 1603 Parnalba-Urban 70.3 16 0 ~ 5 0 9 7.2 brium among the four income groups, though the group between one-half and one minimum wage has a slightly greater relative weight. In the Northeast, however, this equilibrium gives way to an asymmetrical distribution with a heavy concentration in the lowest bracket. Tn Recife, 42.9 percent of the families sampled fell into the first income category; an even Are striking picture is pre- sented by Parnaiba, where 70.3 percent of the families sailed fell within the per capita income bracket defined as up to half the minimum wage. These data reflect the variations in economic development Anteing the nine contexts descr ibed in the Appendix. With regard to Recife, as pointed out in the Appendix, the city's labor force has been characterized over recant decades by intense f luctuations between employment, unem ployment, odd jobs, and various kinds of urban underem- ployment. The evolutionary rate for hiring of employees and dismissals in the metropolitan region of Recife for the period 1978-79 gives some idea of this fluctuation: in industry, the number of workers hired fell from 170 to 166, while the nether of those dismissed rose from 150 to 194; in the service branch, while there was a rise in hirings from 118 to 128, dismissals also rose, from 113 to 125 (FIDEPE, 1980). Moreover, there can be no doubt about the underpayment of the labor force as a determinant of the poverty that dominates the Northeast of Brazil; even in the metropolitan region, 24 percent of all oc- cupied persons work 56 hours or more per week. As will be remembered, the urban economy of Parnaiba is linked to a subsistence- or peasant-based economy cen- tered on the large estates {latifundia), and mercantile or commodity relations have penetrated very little. The town is a collection and distribution center for products

198 of plant extraction. It also has a small manufacturing branch that employs a. very limited number of people-- around 18 percent of the local labor force In full-time employment, while most of its production is put out to domestic or self-employed workers. As regards commerce, here, too, 95 percent c)f active workers are self-employed. Income from these activities is too low for subsistence, and it thus becomes necessary to combine extractive, industrial, and commercial activities and handicraf ts . The bottom line of Table 75 clearly reelects this picture, characterized by a small contingent of families integrated into a formal market, with the majority engaged in oc- casional or irregular activities. An analysis of female fertility for each of the f ive urban contexts and the four income groups yields some interesting observations. Table 76 shows the average number of children born to ever-married women IS years of age or over who, at the time of the survey, came within a given income bracket. This table shows, f irst, that in all the contexts sur- veyed, fertility decreases as per capita monthly income rises. In Santa Cruz do Sul, it makes little difference whether a woman belongs to the first or second income bracket; the greatest decrease is for families with per capita monthly income of over one minimum wage. In Sao Jose dos Campos ~ virtually the same picture holds true, although the coloring is somewhat more vivid. In Recife, however, it does make a difference in the average number of children whether a woman's family has ~ monthly income of between one-half and one minimum wage: compared with the first bracket, the decrease in fertility is around 30 percent. Unfortunately, it was impossible to assess the evolution of fertility for all four income groups in Par- naiba because of the extremely low number of women in the two higher brackets; for the two lower groups, however, there was a drop of 24 percent in the average number of children. Second, the table shows that, even when a given income category is kept fixed, the various contexts show differ- ent levels of fertility: Santa Cruz do Sul and Parnaiba are at one extreme of a slope, especially as regards ache first bracket, while Recife, Cachoeiro do Itapemirim, and Sao Jose dos Cantos, with some small discrepancies among themselves, form an intermediate group. This variation in the average number of children per woman, which reaches its peak--2. 49--for the poorest families, drops in a uni- form manner until it reaches 0.32 for the bracket over

199 . - l N ·. ~ in Sit ~ ~ X O S O ~ C) _O I 4¢ ~ . - £ o :~ O O - £ ~3 - .,, — O o of ~1 C: O a: ED ~ P. At; En P4 - _ c~ ~o ~ ~ 1 _ In Us Us ~ 1 - t_ ~ ~ us t _t ~ 1 . _ ~ Cal sit _ _ _ _ C ~ ~ _ It) _ Cal __ ~ C, :c _ ~ _ I. a: _ V —m ~ O o ~ ~ ~ - O ~ C ~ ~ me al En : · ~ c ~ ~ c - · ~ x: '~ ~ - ~ - ~: x o c :D ~ 0 1` ~ 1 C~ ~1 c~ ~ 1 t - 4 ~ r_ ~o ~ N O · 0 · e I 0 C~ ~ O ~P ~ 1 0 o ~ 1 C~ C~ U' 0` 0 · · ~ 0 · e C~ r. ~ ~ co ~r · ~ · · ~ — tQ o :' _ e ~ tQ P: o ~4 . - 83 o c u C) o o u Ll - 0 m Q ~ O — ~ ~ _ O ~ X ~n ~ x:

200 twice the minimum wage. If Parnaiba is eliminated from the comparison tsince the reduction in the variation may be influenced by the fact that Parnaiba is not represented in the two higher income brackets ), the maximum dif ference between the contexts for the four brackets would then be, respectively, 1. 6G, 1.27, 1. 04, and O .32. In other words, the decrease in the vat iation between contexts as the per capita monthly income of the families rises is a persis- tent trend. This trend may well reflect, among other things, the effect of the population ° s degree of involvement in the f ormal labor market O Such involvement requires more skill, reflected, for example, in a higher level of schooling; it also exposes people to a mass of information related to health, hygiene, sex, and reproduction. In Sao Jose dos Campos, for example' a survey of the big companies, mainly the multinationals, showed that social workers within the companies present the idea of planning to employees whose productivity is falling off; this plan- ning includes the reduction of fertility. ~ ~ ~ In their study of social institutions and reproductive behavior, Loyola and Quinte~ro (1982:43) make the following point: · ~ . social institutions were observed to act basically along Centralist lines, ice., they induce or transmit, in an explicit or diffuse man- ner, the pattern of the small conjugal family, whose corollary is the idea of birth controls According to the viewpoint of the institutional agents, such a reproductive pattern is associated -- again at all points of the survey in general -- to economic and social problems (poverty, cost of living, social marginality, etc.) and appears as a solution offered to such problems in the shout or long term. However, as Loyola and Quintero note further (pp. 43-45), this institutional role varies significantly from one institution and from one context to another. Thus, to give one example of a contextual variation, the most general references to social problems can, in a large town in the South undergoing intense industrialization and with full employment (the case of Sao Jose dos Campos at the Age of the survey), be translated into other, more specific references which basically concern the disorganizing effects on the social fabric of the excessively intense pace

201 of industrialization and immigration; in Parnaiba or Recife, they may take on the connotation of back- wardness as compared with the South, while situa- tions of unemployment and underemployment may be seen as non-residual realities which affect the population as a whole through their pathological consequences: abandoned minors, prostitution (Parnaiba and Recite), marginality and criminality (Recife). Finally, it should be noted that, in addition to the basic criteria used to select the NIHR urban contextso-the prevailing form of organization of production and the social division of labor--or perhaps even as a result of these criteria, the contexts have specific features that may help to explain fertility levels. Thus in 1950, Santa Cruz do Sul, a region with a large population of German origin, already had a fertility rate considered low by Brazilian standards; moreover, family size in Santa Cruz has been highly influenced by property size, which has fallen gradually over the last few decades because of the successive sharing of inheritances among surviving children.

CHAPTER 1 1 CONCLUS IONS DECOMPOSITION OF GENERAL FERTILITY RATES The contribution of changes in age structure, marital status, and marital fertility to declines in general fer- tility rates in 1970-75 is shown for the nine contexts in Table 770 The standardization procedure used (see United Nations, 19791, taking 1970 as the base year , shows that the three factors play specif ic roles in reducing fertil- ity in the different contexts and that distinct interac- tion effects are also present. As can be seen f over the per iod considered, the decline in the general fertility rate varied a great deal from one context to another, as has already been observed with regard to total fertility rates. The largest fall, in Sertaozinho, was around 67.43 per 1, 000 (174.70 -107. 27) At the other extreme was Santa Cruz-Urban ( 90. 41-83 0 7S), with 6~65 per 1,000. Cachoeiro de Itapemir~m, Sao Jose dos Campos (154049-140.20), Recife (174e85~162~00) ~ and Conceicso do Araguaia (236~26-225~70) showed similar declines, which varied from 11 to IS per 1,000; Parnaiba- Urban (227.94-18S.12) and Santa Cruz-Rural (190.48-149.28) had larger declines of around 40 per 1, 000. The only exception was Parnaiba-Rural (250.49-273.26), with a growth of 22.27 per 1,000 between 1970 and 1975. In all the contexts where there was a reduction in the general fertility rate, except for Santa Cruz-Urban, the decline in marital fertility was the most directly responsible factor . In Cachoeiro, Recife, Sao Jose, and Araguaia, the decline can be totally explained by the reduction in marital fertility, and would have been even larger if age structure and mar ital status had not con- tributed to the attenuation of this reduction. In Cacho- eiro, Recife, and Araguaia, the contribution of marital . 202

203 ._, so as: U] V) .,' as: sol :: t) :, U) ~s: o a) :' Q a, ~o ·, 1~ N ·,1 ·~4 ·. , O ' ~ ~n QJ ~ ~ X · - o U] c) C: ~o ·~ =: Z C) ._, _ o c .... N a 4, U] ~ C C L. ~ _ C ~ :1 U) Cl d, 0 ~ h4 L4 _ _ 0 ~ ~ o. U o 0 ~ ~ 0 U: - 3: o o U— _ ~o U C a e i~ ~ C C ~n ·. Ec] -~ JJ ~ a, E~ ~ ~n ~ ~ ~ ~ r~ u~ D t_ ~ ul , · · · · · · · . ~4 ~ ~4 ~ ~ ~ _d I _ t ~ ~4 1 ~4 O4 U~ In ~ ~ ~ ~ ~ ~o . · . . · · . . . _' ~ O D O O ~ t— ~4 \0 ~ \0 1 1 1 1 a' ~ ~ ~ ~ <r' ~D O ~ `0 ~ _ O 0` ~ ~ ~ e · · · · 0 ~ ~ ~ r~ _~ 1 ~ _1 1 1 1 1 1 r" · ~ cr' _. 1 —4 ~ r~ 0 ~o o U' · · · . . . O O ~ 1 1 -4 `0 O c~ 1 1 O U~ `0 ~ `0 · · . O O O 06 cr' ~ · ~ 1 1 r~ ~ ~ ~ - 4 0 ~ o' U~ (D . u~ C~ ~ ~ · · . ~ · · . . . ~ ~ ~ ~ ~ ~ O _~ r~ 1 1 1 ~ 1 1 1 u~ ~n ~ u~ ~o ~ t0 r~ ~ ~ ~o · · · · · · ~ · - ~r ~ ,l~ ~ ~ 0 0 r~ ~ —~ c~ 1 1 1 _~ _l 1 1 1 ~ ~ ~ \0 `0 ~ \0 ~ \0 C~ ~ ~ 0 ~0 U~ ~ ~ ~ . · . · · · · · ~ ~4 ~ U~ O ~ O O _~ ~ 1 1 ~ _1 _ 1 1 1 ~ O 10 O · ~ . ~ U' ~ ~ O 1 1 1 1 C~4 ~ · ~ O 1 C ~ ~ ~ C ~ C ~ ~ ~4 _ _ W 0 4, :, : C, U 1 _ ~ ~ 4` he ~ >. ~ V' ~ .e _ _ 4, _ _ _. .4 _ O 48 0 ~ O ~ ~ ~ c ~ e w c ~ ,` ~ ~ 0 0 0 U ~ .4 — - 4 _ ~ _ ~ ~ ~ V Cl L. ~ - 4 U ~ C' ~ U ~ ,., ~ ~ ~ _ ~ _ ~ _ Ll _ _ ~ ~ ~ ~ ~ ~ ~ s ~ e C c :' >' 4'— ~4 4 — W O ~ W oc ~ _ 4 U— C ~ 0. O ~ · ~ 0 \0 1 C 41 C O _ e 0 C 41 X 'uC 4 Ll o E. ~0 ~0 C C C ~ ~C O o E.

204 status to an increase in the general fertility rate sur- passed that of the age structure of the female population; in Sao Jose, the opposite was the case. It is interesting to note that in Parnaiba-Urban and Santa Cruz-Rural, even though, as already stressed, the major factor determining the decline was the fall in mari- tal fertility, the change in the distribution of the female population according to marital status also con- tributed to the decline. The age structure of the female population, on the other hand f showed little change in the same period' thus favoring an i n - ~:~^ i ~ she ~~_ fertility rate. ~ by_ ~ ^~4 —~~= ~~11=L ~ ~ In sertaozinho, it was not only the fall in marital fertility, but also the change in the age structure of the population that contributed to the decline in the general fertility rate. In Santa Cruz-Urban, these two factors were almost equal in contributing to the small decline for that region. Finally, the increase in the general fertility rate shown for Parnaiba-Rural was the result of an increase in the marital fertility rate, a more favorable distribution by marital status, and an age distribution which favored a drop in fertility. DECOMPOSITION OF TOTAL FERTILITY As seen above, Bongsarts' model for estimating the total fertility rate is expressed by the relation TER ~ 15.3 x Ci x Cat x Ca x Cat tl). In other words, the starting total fecundity rate of 15.3 is gradually reduced by the action of the various inhibit- ing factors. The first of these is postpartum infecun- dity, Ci, which when applied to 15.3 reduces it to the total natural marital fertility rate: TNMFR ~ 15.3 x Ci. This in turn, when reduced through the ef feats of Cc and Ca, becomes the total marital fertility rate, or TMFR ~ TNMFR x ~ x Ca. Finally, the latter, when subjected to the effect of Cm, becomes the total fertility rate, or

205 TFR = TMFR x Cm. Table 78 shows the decomposed values for the different contexts studied. As Man be seen, the values of C i in Cachoeiro do Itapemirim and Conceicao do Araguaia caused the greatest reduction--around 15 percent--in the total fecundity rate owing to the ef feet of lactation. In Recife, this effect was the lowest of all--only 6 percent. The prevalence of deliberate control of marital fertility was undeniably the main f actor accounting for the reduc- tion in the natural marital fertility rate in all con- texts. In Cachoeiro and Sao Jose dos Campos, this effect reached 70 percent, followed by Santa Cruz-Urban and Ser- taozinho, where it was 65 percent. In Santa Cruz-Rural and Recife, it accounted for 61 percent and 57 percent, respectively, of the fall in the rate in question. The rate for Parnaiba-Rural was reduced by only 24 percent. It should be stressed that in all contexts, although in a - differentiated manner, contraception was the factor directly responsible for the difference between the total natural marital and total marital fertility rates. As for the effect of the absence of marriages, it was very small in Parniaba-Rural and Sertaozinho, varying between 16 and 17 percent. In Parnaiba-Urban and Conceicso do Araguaia, this effect was about 22 percent, increasing to about 30 percent for Santa Cruz. In Sao Jose dos Carepos and Re<:ife 34 percent of the women were not yet married, and in Cachoeiro it was 35 percent, bringing the total marital fertility rate down f rom 3 . 91 to 2. 53 . In the other f ive contexts, this effect was practically uniform--between 21 percent and 24 percent. F INDINGS AT THE NATIONAL/STATE: AND LOCAL I~EVELS When the f indings in Parts I and IT of this report are compared, they are found to be complementary and mutually supportive. Both data sets clearly show an accelerated fertility decline in Brazil during the 1970s. Both indi- cate the importance of consensual unions in Brazilian nuptiality patterns, as well as the cliff iculties involved in gather ing data on these unions; they also point to shif ts in age at mart iage e However, the national- and local-level analyses lead to the same conclusion: that the primary factor responsible for Brazil's accelerated fertility decline is declining marital fertility. This decline can in turn be traced, at both levels of analysis,

206 ~3 - C L4 ~ a: ,. N oo U~ ca X o C) · - z ·,. A ·.-o LJ a, o E~ o ~q · - . - X o a. a, ~3 E~ o o ~ . · - U C o C) a . ~ ~ C C ~ ~4 ~d - u ~D ~: CJ C ~ :' ~o o c - o o a. o ~ CJ o o ~n ~8 3 C, 4, C C ~Q :) o q:S O — _ ~ O ~ ~ a. U ~ C., — <. S ~ r. ~ r~ a. ~o ~ O ~ ~ 3) ~ o' r~ ~ 51 ~o . · . . . · · - U~ O ~ :3 0 0 0 _ _ _ 0 ~ ~ r~ a, ~ ~ ~ ~ D O ~ `0 D ~— D . · . . · · . ~ U~ O ~ O O ~ O ~4 ~4 · . U' O ~0 O' · ~ O ~4 r~ r. CD . · . U~ O ~4 _ r. 0` CO · . U~ O - · . ~ O d r~ · — ~ O o 0` · . O >. 4,' — U >' V C C U V ~ 04 ~J D O O E. ~ tD u~ · ~ O ~0 ~n O · ~ O O ~ U' ~ ~ O · 10 ~ · · . . O U1 0 0\ ~ · ~ O O O ~ ~ 40 (D O · . — ~ O `8 ~ ~ 0 ~0 \0 · · . ~ O U~ O r~ O · · — 0 0 ~n o ~ U' · . O < - ~ — O ~ ~ ~ tD ~n · . · · . ~ O O ~ O ~4 >q - - _ v {S: - - ~ C Z O - _ O 0 4 ~ e c ~ O r~,, _ v u o - - C) L CJ 0 4 - S v o e 50 -

207 to changing patterns of contraceptive use: increased prevalence, coupled with the use of more ef festive methods (the pill and sterilizatiOn). Among the other proximate determinants of marital fertility, postpartum infecund- ability has relatively little impact in Brazil, where the practice of breastfeeding is quite limited; abortion, shown in both data sets to be increasing, is nevertheless an indeterminate f actor because of the usual data-gather- ing problems. Finally, both analyses point to a combina- tion of factors--changes in relative income, increased educational attainment, and increased female labor force participation--as the main socioeconomic determinant of Brazil' s changing patterns of contraceptive use: these factors are associated with increased desires for consumer durables, the diffusion of information on family-size limitations and therefore an increase in conscious fer- tility control.