Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

This chapter offers a general overview of the often spectacular rise of the share of cohabitation in the process of union formation in 24 Latin American and Caribbean countries during the last 30 years of the twentieth and the first decade of the twenty-first century. Firstly, a brief ethnographic and historical sketch will be offered with the aim of illustrating the special position of many Latin American regions and sub-populations with respect to forms of partnership formation other than classic marriage. Secondly, the national trends in the rising share of cohabitation in union formation will be presented for men and women for the age groups 25–29 and 30–34. This is extended to full cohort profiles covering all ages in Brazil and Mexico. Thirdly, we shall inspect the education and social class differentials by presenting the cross-sectional gradients over time. The fourth section is devoted to the framework of the “second demographic transition” and hence to the de-stigmatization of a number of other behaviors that were equally subject to strong normative restrictions in the past (e.g. divorce, abortion, homosexuality, suicide and euthanasia). The last section deals with the household and family contexts of married persons and cohabitors respectively.

The chapter is not only meant to offer a statistical description, but also to raise several points that should facilitate an interpretation of the phenomenon of the “cohabitation boom”. A short introduction of the issues involved is now being presented.

In many provinces, and especially those with larger native and black populations, cohabitation and visiting unions have always existed as alternatives to the classic “European” marriage. However, as the data from up to five census rounds indicate, the rise in cohabitation occurred both in such areas with “old cohabitation” practices and in those where cohabitation had remained much more exceptional till the 1970s. In other words, there is now a sizeable amount of “new cohabitation” besides or on top of “old cohabitation” (see also: Castro-Martín 2002; Binstock 2008).

The same census data also document the existence of a universal negative cohabitation- education gradient, with women with higher levels of education cohabiting less and moving into marriage in greater proportions. The existence of a negative gradient with education, and by extension also by social class, is commonly interpreted as the manifestation of a “pattern of disadvantage”. In this pattern, the poorer segments of the population would not be able to afford a wedding and the setting up of a more elaborate residence, but they would move into other forms of partnership such as cohabitation or visiting unions. In this view, “cohabitation is the poor man’s marriage”. The “crisis hypothesis” follows a similar line of reasoning. Given the deep economic crises and spells of hyperinflation during the 1980s in almost all Latin American countries, the lower social strata would have reacted by further abandoning marriage and resorting to more cohabitation instead.

The matter is, however, far more complicated than just sketched. Given this negative cross-sectional gradient with education, one would expect that with advancing education over time many more persons would get married rather than cohabiting. The advancement in male and female education in Latin America has been very pronounced since the 1970s, and yet, just the opposite trend in marriage and cohabitation is observed compared to the one predicted on the basis of the cross-sectional education gradient: there is now far more cohabitation and much less marriage. In other words, the changing educational composition not only failed to produce a “marriage boom”, but a “cohabitation boom” developed instead. This not only reveals once more the fallacy inherent in the extrapolation of cross-sectional differentials, but illustrates even more strongly that other factors favorable to cohabitation must have been “flying under the radar”. In this chapter we shall therefore also explore to what extent ideational factors, especially in the domains of ethics, sexuality, secularization and gender relations, could have contributed to the emergence of the “cohabitation boom”. This brings us inevitably to the issue of a possible partial convergence of several Latin American populations to the pattern of the “Second Demographic Transition” (SDT) (Lesthaeghe and Van de Kaa 1986; Lesthaeghe 2010).

The rise in cohabitation also begs the question whether cohabiting persons form nuclear families or stay with their own parents or kin instead and hence continue to rely on residential extended family structures. In other words, is the rise of cohabitation a source of family simplification (nuclearization), or are the residential household compositions essentially untouched?

We shall now turn to the details of the points just sketched above.

2 “Old” and “New” Cohabitation

Native and black populations in Latin America and the Caribbean have been known to have maintained patterns of union formation other than classic marriage. (e.g. Smith 1956; Roberts and Sinclair 1978). In the instance of American Indian indigenous populations, ethnographic evidence shows that they did not adhere to the group of populations with diverging devolution of property through women. As argued by J. Goody (1976), populations that pass on property via a dowry or an inheritance for daughters (i.e. populations with “diverging devolution” of family property via women) tend to stress premarital chastity, control union formation via arranged marriages, elaborate marriage ceremonies, and reduce the status of a married woman within the husband’s patriarchal household. Moreover they tend toward endogamous marriage (cross-cousin preference) or to caste or social class homogamy. Through these mechanisms the property “alienated” by daughters can still stay within the same lineage or clan or circulate within the same caste or social class. Populations that are hunter-gatherers or who practice agriculture on common community land, have fewer private possessions, no diverging devolution of property via dowries, no strict marriage arrangements or strict rules regarding premarital or extramarital sex. Instead, they tend to be more commonly polygamous with either polygyny or polyandry, have bride service or bride price instead of dowries, and practice levirate or even wife-lending. The dominance of the latter system among American natives can be gleaned from the materials brought together in Table 2.1.

Table 2.1 Distribution of 51 ethnic populations according to selected characteristics of their marriages and sexual unions
Full size table

Table 2.1 was constructed on the basis of the 31 ethnic group references contained and coded in the G.P. Murdock and D.R. White “Ethnographic Atlas” (1969), and another 20 group specific descriptions gathered in the “Yale Human Areas Relation Files” (eHRAF 2010). Via these materials, which refer mainly to the first half of the twentieth century, we could group the various populations in broader ethnic clusters and geographical locations, and check the presence or absence of several distinguishing features of social organization.

Of the 41 native groups mentioned in these ethnographic samples, only one had an almost exclusively monogamous marriage pattern, whereas the others combined monogamy with polyandry often based on wife-lending, occasional polygyny associated with life cycle phases (e.g. associated with levirate), more common polygyny, or serial polygyny in the form of successive visiting unions. For 26 native Indian groups we have also information concerning the incidence of extramarital sex or of visiting unions. In only six of them these features were rare. Furthermore, vnone have a dowry, which implies that the feature of diverging devolution is absent, and that, compared to their European colonizers, these populations are located on the other side of the “Goody divide”. As expected, they have the opposite pattern in which the prospective groom or the new husband has to render services to his in-laws or pay a certain sum of money to his wife’s kin. In a number of instances, there was also a custom of women or sister exchange in marriage between two bands or clans, and there were also instances with just gift exchanges or no exchanges at all. And finally, mentions of elaborate marriage ceremonies were only found among the references to Mexican or Central American indigenous groups, whereas the others had marriages with a simple ritual only, and often had a “marriage” as a gradual process rather than a single event.

The data presented in Table 2.1, however, essentially refer to smaller and more isolated indigenous populations who had maintained their lifestyles until the beginning of the twentieth century, and as a consequence they constitute a selective sample. At the time of the European conquests during the sixteenth century also large states existed (e.g. Aztec, Maya, Inca), which were both highly centralized and “ritualized”. These features facilitated the conversion to Christianity, and hence the adoption of a monogamous Christian marriage. By contrast, nomadic tribes and small indigenous populations in isolated places such as mountain canyons or the forest could maintain their traditions much longer and resist both, economic and administrative control from the center and the adoption of Christianity. These dualities help to explain the diverging historical tracks followed by indigenous populations. Furthermore, also the “mestization” of large numbers of them and the concentration of these populations in larger villages or around agricultural enterprises fostered conversion to Catholicism and the adoption of the Christian marriage pattern.

The story for the New World black and mixed populations is of course very different, since these populations were imported as slaves. As such they had to undergo the rules set by their European masters, or, when freed or eloped, they had to “reinvent” their own rules. When still in slavery, marriages and even unions were not encouraged by the white masters, given the lower labor productivity of pregnant women and mothers. And for as long as new imports remained cheap, there was little interest on the part of the owners in the natural growth of the estates’ slave population. The “reinvented” family patterns among eloped or freed black populations were often believed to be “African”, but in reality there are no instances where the distinct West African kinship patterns and concomitant patterns of social organization are reproduced (strict exogamy, widespread gerontocratic polygyny). Instead, there is a dominance of visiting unions, in which the woman only accepts a male partner for as long as he contributes financially or in kind to the household expenditures and where the children of successive partners stay with their mother. Not surprisingly, diverging devolution is equally absent among the New World black and mixed populations reviewed by our two ethnographic samples. In this regard, they do follow the pattern of West-African non-Islamized populations.

The white colonial settler population or the upper social class by contrast adhered to the principles of the European marriage (“Spanish marriage”, “Portuguese nobres marriage”) being monogamous, based on diverging devolution and hence with social class as well as preferred families endogamy. However, this European pattern was complemented with rather widespread concubinage, either with lower social class women or slaves (see for instance Borges 1985 and Beierle 1999; for the Bahia colonial upper class in Brazil and Twinam 1999; for several Spanish speaking populations). Children from such unions in Brazil could easily be legitimized by their fathers via a simple notary act (Borges 1985).

As indicated, the data of Table 2.1 should of course be taken as an illustration, and not as an exhaustive classification of Latin American ethnic populations. But, in our opinion, they clearly demonstrate that “marriage” as Eurasian societies know it, initially must have been a fairly irrelevant construct to both indigenous and New World black populations, and subsequently, just an ideal or a formal marker of social success.

So far, we have mainly dealt with the historical roots of the diverse patterns of union formation. But more needs to be said about the influence of institutional factors and immigration.

The Catholic church and the states generally tended to favor the “European” marriage pattern, but originally with quite some ambiguity. First, the Catholic clergy, and especially those in more distant parishes, did not observe the celibacy requirement that strictly. Second, many Christian and pre-Colombian practices were merged into highly syncretic devotions. The promotion of the Christian marriage was mainly the work of the religious orders (Franciscans, Augustinians, Dominicans, and until the end of the eighteenth century also the Jesuits). At present, that promotion is vigorously carried out by the new Evangelical churches which have been springing up all over the continent since the 1950s, and most visibly in Brazil and Peru.

Also the role of the various states is often highly ambiguous. Generally, states copied the European legislations of the colonizing nations and hence “officially” promoted the classic European marriage, but more often than not this was accompanied by amendments that involved the recognition of consensual unions as a form of common law marriage and also of equal inheritance rights for children born in such unions. In Brazil, for instance, Portuguese law had already spelled out two types of family regulations as early as the sixteenth century (Philippine Code of 1603), namely laws pertaining to the property of notables (nobres) who married in church and transmitted significant property, and laws pertaining to the countryfolk (peões) who did not necessarily marry and continued to live in consensual unions (Borges 1985). Furthermore, it should also be stressed that many central governments were often far too weak to implement any consistent policy in favor of the European marriage pattern. Add to that the remoteness of many settlements and the lack of interest of local administrations to enforce the centrally enacted legislation.

However, as pointed out by Quilodrán (1999), it would be a major simplification to assume that this “old cohabitation” was a uniform trait in Latin American countries. Quite the opposite is true. In many areas, late nineteenth century and twentieth century mass European immigration (Spanish, Portuguese, Italian, German) to the emerging urban and industrial centers of the continent reintroduced the typical Western European marriage pattern with monogamy, highly institutionally regulated marriage, condemnation of illegitimacy and low divorce. As a consequence, the European model was reinforced to a considerable extent and became part and parcel of the urban process of embourgeoisement. It is interesting to note that even the Communist party in Cuba initially wanted to promote classic European-style marriages. To this end, they considered erecting “marriage palaces” and organizing group marriages, so that also poorer people would be able to celebrate the event “with all the luxuries of a bourgeois wedding” (Martínez-Allier 1989: 140).

The combination of the various factors just outlined not only caused the incidence of cohabitation to vary widely geographically and in function of the ethnic mix, but also produced the emergence of a marked gradient by educational level and social class: the higher the level of education, the lower the incidence of cohabitation and the higher that of marriage. This negative cohabitation-education gradient is obviously essentially the result of historical developments and long term forces, and, as we shall illustrate shortly, found in every single one of the countries studied here. The gradient is not the outcome of a particular economic crisis or decade of stagnation (e.g. the 1980s and early 1990s).

3 The Latin American Cohabitation Boom: The National Trends

Latin American censuses have historically provided an explicit category for consensual unions (uniones libres, uniones consensuales). The examination of the questionnaires of all Latin American and Caribbean censuses conducted between the1960s and 2000s reveals that in the vast majority of them cohabitants could be explicitly indentified either through the variables ‘marital status’ (dominant approach) or ‘union status’ (quite common in Caribbean countries) or through a direct question (e.g. Brazil and recently in Argentina and Surinam). A methodological problem emerges, however, when individuals that cohabited in the past and were no longer in union at the time of the census report themselves as singles (Esteve et al. 2011). This clearly exaggerates the proportion of singles and affects the ratio between married and cohabitating couples as we observe ages that are increasingly distant from those in which union formation was more intense. To minimize bias, our analysis focuses on young ages, mainly 25–29.Footnote 1 However, cohabitation may not be an enduring state and subsequent transitions to marriage are often the rule. In such circumstances, those with early entries into a partnership may already be in the process of moving from cohabitation into marriage at ages 25–29, whereas those with later partnering, such as the more educated, may still be in the process of moving from singlehood to cohabitation (Ni Brolchain and Beaujouan 2013). In that instance there would be a bias in favor of marriage for the less educated and in favor of cohabitation for those with longer educational careers. In the Latin American setting there is simply no increase in the proportions married in any of the education groups at any age, and hence this timing effect of entry into a partnership barely affects the outcomes that will be described. This is furthermore confirmed by inspecting the share of cohabitation in the next age group 30–34 and by following men aged 25–29 and 30–34 as well. In other words, the “quantum” effect (i.e. the sheer size of the ubiquitous rise in cohabitation) by far outweighs any tempo-related distortion.

Several researchers (e.g. Ruiz Salguero and Rodríguez Vignoli 2011; Rosero Bixby et al. 2009; López-Ruiz et al. 2008; Rodríguez Vignoli 2005; García and Rojas 2002) have used census data to explore cohabitation patterns in Latin America. Some of them did so on the basis of the Integrated Public Use Microdata Series (IPUMS) that have been collected and harmonized at the University of Minnesota Population Studies Center (Minnesota Population Center 2014). Also, estimates of the share of consensual unions among all unions were made by the US Census Bureau (2004) for the censuses of the 1950s and 1960s in a more limited number of countries.

Previous research reveals a remarkable rise of the share of consensual unions among all unions, and this rise most probably already starts during the 1960s in a number of countries (Fussell and Palloni 2004), involving both countries with an initially very low incidence of cohabitation and countries with higher levels. The early cohabitation shares reported by Fussell and Palloni pertain to the unions of women aged 20–29. These data indicate that Argentina (5.8 % cohabitation of all unions in 1950), Uruguay (5.7 % in 1960), Chile (3.0 % in 1970) and Brazil (5.1 % in 1960) belong to the former category. Peru (20.9 % in 1960) and Colombia (13.5 % in 1960) are typical examples of the latter group with later rises. However, countries with pre-existing high levels of what we have called “old cohabitation” did not witness the onset of such a trend until much later. Examples thereof are Guatemala (56.1 % in 1950) or Venezuela (29.7 % in 1950), the Dominican Republic (44.4 % in 1960) or El Salvador (34.2 % in 1960).

The results that will be reported from here onward stem from the extensive analysis of the harmonized Latin American census microdata samples available at IPUMS international (Minnesota Population Center 2014).This analysis uses as many census rounds between 1970 and 2010 as possible (see Appendix Table 2.8). Consequently, with the exception of few areas, the time series generally capture the initial rises of the share of cohabitation. The results are shown in Table 2.2 for 24 countries, and for men and women aged 25–29 and 30–34 respectively.

Table 2.2 Percent cohabiting among all persons in a union (married + cohabiting), 25–34, by sex and census round, Latin America and the Caribbean, 1970–2010
Full size table

The data in Table 2.2 not only document the marked heterogeneity of Latin American countries at the onset, but also the acceleration in an already upward trend during the 1990s. There are essentially two groups of countries, i.e. countries that had a strong tradition of marriage with little cohabitation to start with, and countries in which cohabitation was more widespread and had stronger historical roots.

During the early 1960s (1970 census round) the share of cohabitation among all men or women 25–29 in a union varied between about 5 and 20 % in countries with low levels of “old cohabitation”, i.e. Argentina, Brazil, Chile, Colombia, Costa Rica, Mexico, Puerto Rico, and Uruguay. However, a genuine cohabitation boom took place during the 1990s that drove up these percentages to levels between 25 and 70 %. The 1990s were particularly significant for Colombia where the share of cohabitation for women 25–29 jumps from about 20 % in 1973 to almost 50 in 1993 and over 65 in 2005. Less spectacular, but equally noteworthy are the large increments in Argentina and Brazil where the cohabitation shares initially remained fairly stable around 15 %, but then increased during the 1990s by about 30 percentage points compared to the 1970 figure. Increments over that period of about 20 percentage points are witnessed in Costa Rica and Chile. But the “late starters” are Mexico, Puerto Rico, Chile, Paraguay and Uruguay with only modest rises till 2000.

The first decade of the twenty-first century is characterized by several further spectacular rises in the initially “low” group of countries. The latest census figures for the 2010 round indicate that the share of cohabitation passed the 50 % threshold in Brazil and Costa Rica, and that even the 60 % mark was amply passed in Argentina, Colombia, and Uruguay. For Puerto Rico and Chile we have no 2010 data, but Mexico, the other late starter, was clearly catching up and coming close to a cohabitation share of 40 %.

Among the countries with about 30 % or more cohabitors among women or men 25–29 in unions in the 1970s census round, i.e. among those with sizeable categories of “old cohabitation”, there are also remarkable rises that took place during the last two decades. Clear examples thereof are the Dominican Republic, Ecuador, Venezuela, Peru and even Panama which had the highest levels to start with in 1970.

For the remaining countries in Table 2.2 we have only one or two points of measurement, but according to the 2000 census round, most of them had a cohabitation share in excess of 35 % and up to about 60 % (highest: Cuba, Jamaica, Honduras, Nicaragua). Furthermore it should be noticed that several Central American countries tend to exhibit a status quo, but at high levels. This holds for Guatemala, El Salvador and Nicaragua, but as indicated above, not for Costa Rica and Panama where the upward trend was continued.

Judging from the most recent 2000 or 2010 figures, cohabitation has overtaken marriage among men 25–29 in 16 of the 23 countries (no data for men in Trinidad and Tobago), and among women 25–29 in 13 of the 24 countries considered here. In 1970 there was only one case (Panama) among 12 countries with a cohabitation share in excess of 50 %, and in 1980 there were only 2 (Dominican Republic and Panama) among 13 countries.

Finally, it should also be noted that the figures for the next age group, i.e. 30–34, are roughly 10–15 percentage points lower. There are two competing explanations for this feature. First, the drop off could be due to the post-cohabitation transition into marriage, and this would be indicative of cohabitation being only a transient state as in several European countries. Alternatively, it can be explained by a cohort effect with the older generation having experienced less cohabitation when they were in their late twenties. This explanation is particularly likely in periods of rapid change. In this instance cohort profiles should be layered horizontally rather than dropping off with age, meaning that each generation climbs a step further upward with respect to the incidence of cohabitation. This would, furthermore be indicative of cohabitation being a much more permanent state over the life cycle of individuals. Note, however, that such stability of cohabitation over age and time does not imply stability with the same partner.

The availability of several successive censuses permits the reconstruction of the cohort profiles stretching over the entire adult life span. It should be noted, however, that this is a reconstruction at the macro level, and that no individual transitions are recorded (a life table analysis of individual cohabitation durations would then be needed). Nevertheless, the cohort profiles are still very instructive, as can be seen from the reconstructions for Brazil and Mexico in Fig. 2.1.

Fig. 2.1
figure 1

Age distributions of the share of cohabitation for all women in a union and corresponding cohort profiles (C.). Brazil and Mexico, 1960–2010

Source: Authors’ elaboration based on census samples from IPUMS-International

Full size image

The Brazilian age distributions of the share of cohabitants among all partnered women are dramatically moving up at all ages during the window of observation between 1960 and 2010. For all cohorts up to the one born in 1980, this results in flat rather than downward slopes of cohort profiles starting at age 20, and the gap between the successive generations also widens with the arrival of the younger ones born between 1960 and 1980. All of this is illustrative of a very clear generation driven pattern of social change, with cohabitation being a much more enduring state over long periods in the life cycle. In other words, cohabitation is not just a matter of a short spell of partnership trial(s) but more like a marriage substitute. The slightly upward slopes for the older cohorts may also be indicative of older women moving into cohabitation following a marriage interruption due to divorce or widowhood. The cohort born in 1980, by contrast, shows the downward slope which is normally associated with greater fractions moving from cohabitation to marriage. For this younger Brazilian cohort, which starts at a much higher level of cohabitation in their early twenties than their predecessors, there may still be some shift associated with a pattern of “trial marriage” going on.

The Mexican data for the earlier censuses are based on a one percent sample only, which explains their bumpier patterns. This, however, does not affect the basic interpretation of what happened. Firstly, Mexico’s later take-off is very clearly in evidence with the initial cohort lines being fairly undifferentiated. The big change comes between 2000 and 2010, when the share of cohabitation increases for all ages, including the older ones. This not only means that the later cohorts born after 1970 become more differentiated, but also that the cohorts born in the 1970s have increasing rather than decreasing percentages cohabiting after the age of 25. Secondly, the same feature is found as for the youngest cohort in Brazil: a downward profile between age 20 and 30. Evidently, also in Mexico, as many more younger women initiate a partnership via cohabitation, a larger segment of them coverts their consensual union into a marriage. However, this movement among the youngest cohort does not at all prevent them from reaching higher levels of cohabitation by age 30.

4 The Education Gradient

We have already pointed out that the negative cross-sectional gradient of cohabitation with rising female education is a historical reflection of ethnic and social class differentials in Latin American and Caribbean countries. This negative slope is found in all countries considered here, and as the data of Fig. 2.2 indicate, this was already clearly in evidence prior to the post-1970 cohabitation boom.

Fig. 2.2
figure 2

Share of cohabitation among all unions of women 25–29 by level of completed education, country and census round (Source: Authors’ elaboration based on census samples from IPUMS-International and National Statistical Offices)

Notes: < Prim Less than Primary Completed, Prim Primary Completed, Sec Secondary Completed, Uni University Completed. Some college is included in university completed in Colombia 1993.There is no category for less than primary in Jamaica 2001. We do not have data on educational attainment for Guatemala 1994, Paraguay 1982–1992 and Peru 1981

Full size image

Taken individually, each of the negative gradients in Fig. 2.2 could be interpreted as the manifestation of the “pattern of disadvantage”. However, given the often spectacular rises since the 1970s, this interpretation would fall considerably short of accurately representing the situation. In fact, in all countries and in all education groups there is such an increase in the share of cohabitation. This obviously includes sometimes dramatic catching up among women with completed secondary and completed university educations. Such increases at the top educational layers obviously cannot be taken as a manifestation of a “pattern of disadvantage”. Clearly, there is a substantial amount of “new cohabitation” that developed on top of the historical “old cohabitation” during the last four decades.

There are, however, substantial differences among the countries represented in Fig. 2.2 Brazil, for instance, is the only country in which the largest rise of the share of cohabitation of partnered women 25–29 is still to be found among women with incomplete primary education. Over the 40 years span, i.e. from 1970 to 2010, Brazilian women with secondary and higher education are the more reluctant ones to swap marriage for cohabitation. This does, however, not stop such women to increase their cohabitation share from virtually zero in 1970 to some 35 % in 2010.

Venezuela comes closest to the Brazilian pattern, but the largest increment is found among women with completed primary education. Also in this country, the catching up of cohabitation among women with completed secondary or higher education is modest, and of the order of 20 percentage points over three decades.

The next group of countries is made up of cases in which the increments are roughly of equal importance in all four education groups. This group comprises Argentina, Chile, Costa Rica and Mexico. These are all countries with overall low levels of cohabitation to start with, but with an original “pattern of disadvantage”. Given similar increments in all groups, this negative gradient is maintained throughout. The Colombian pattern of change over three decades is also quite evenly spread over the various education categories, but the successive increments are much larger than in the previous countries. Moreover, the growth is most pronounced in the middle education categories. Similarly, also Ecuador provides an example with the largest increment for women with completed secondary education, but the overall rise is more modest than in neighboring Colombia. In the other Andean country, Peru, the current pattern of 2007 has become almost flat for the first three education groups at no less than 70 % cohabiting. Women 25–29 with completed tertiary education have crossed the 50 % mark, which was about the level for Peruvian women with no more than primary education in 1993.

The case of Uruguay merits attention in its own right. In 1975, the country also exhibited the classic negative gradient with education, but at low levels for all groups, i.e. not exceeding 20 %. During the next 20 years, the growth was modest and very even. But between 1996 and 2010, a truly spectacular shift occurred from marriage to cohabitation, resulting in an almost flat gradient located at 70 % cohabitation and only 30 % marriage for women 25–29. Among women with completed tertiary education, Uruguay now has the highest percentage cohabiting women 25–29 of all the countries considered here, including the ones with long histories of traditional cohabitation.

The last group of countries is composed of those with long traditions of cohabitation especially among the less educated social classes. These countries are typically in Central America or the Caribbean: Dominican Republic, El Salvador, Nicaragua and Panama. In all four countries the original gradient, measured as of 1970, was very steep, with a share of cohabitation in the 50–90 % range for the lowest education group, and a share not exceeding 12 % for their small group of women with completed university education. In all instances, women with completed secondary education or more have been catching up. In El Salvador, this gain was very modest. In Nicaragua the increment among the middle education groups is already much more pronounced, but this rise occurred essentially between 1971 and 1995, and not so much thereafter. The other two countries in this group with a long cohabitation tradition, i.e. the Dominican Republic and Panama, provide examples of further increments above the initial 70–80 % cohabitation among the least educated women 25–29. This is remarkable given the high levels to start with. However, even more striking is the very substantial catching up in all the other education categories. University educated women 25–29 in both Panama and the Dominican Republic now have an equal 50–50 share of cohabitation and marriage, whereas the middle categories have reached percentages between 70 and 90, i.e. nearly as high as those in the lowest education group.

The upward shifts of the share of cohabitation during the last three or four decades have occurred in tandem with very considerable improvements in education among women in these countries. This can be gleaned from the data in Table 2.3 representing the percentages of all women 25–29 who have completed either full primary or full secondary education. The point here of course is that the group of women with less than complete primary education have become more marginal, and that women with full primary education of today basically belong to the same social strata as those with no or incomplete primary education three decades ago.

Table 2.3 Percentages of women 25–29 with completed primary and completed secondary education by country and census round
Full size table

Considering these major improvements in educational levels described in Table 2.3 in tandem with a negative education gradient for the prevalence of cohabitation, one would project declining overall proportions cohabiting and rising proportions being married. Of course, just the opposite has happened, and quite dramatically so. In other words, the effect of a changing educational composition of the population did not at all work out in the expected direction. Hence, all the changes in cohabitation in Latin America are due to individual changes, and not at all due to the educational composition change.

Now that an explanation based on such a composition shift can be discarded completely, we need to explore other avenues to account for the spectacular rises in cohabitation in all these countries, regions and social strata.

5 Explaining the Rise in Cohabitation

A useful framework for the analysis of any new form of behavior is the “ready, willing and able” (RWA) one used by Coale (1973) to interpret the historical European fertility transition, and elaborated by Lesthaeghe and Vanderhoeft (2001) to accommodate heterogeneity and the time dimension. The “Readiness” condition states that the new form of behavior must have an economic or psychological advantage, and hence refers to the cost-benefit calculus of a particular action compared to its alternatives. The “Willingness” condition, by contrast, refers to the religious and/or ethical legitimacy of the new form of behavior. And the “Ability” condition states that there must be technical and legal means available which permit the realization of that “innovation”. Note, however, that the RWA-conditions must be met jointly before a transition to a new form will take place. It suffices for one condition not being met or lagging for the whole process of change coming to a halt.

In the instance of cohabitation, a number of economic advantages are easily identified. First, compared to legal marriage, cohabitation is an “easy in, easy out” solution. This implies, more specifically, (i) that considerable costs are saved by avoiding more elaborate marriage ceremonies, (ii) that parents and relatives or friends are presented with the outcome of individual partner choice as a fait accompli, and (iii) that the exit costs from cohabitation, both financial and psychological, are considerably lower than in the case of a legal divorce. In other words, cohabitation is the quicker and cheaper road to both sexual partnership and economies of scale. And in many instances, such shorter term advantages may indeed weigh up against the main advantage of marriage, being a firmer longer term commitment.

In addition to these general economic advantages, the rise in cohabitation can also be a response to the economic downturns of the 1980s and the slow recovery of the 1990s. Potential couples in these instances could postpone entry into a union of any type. Alternatively they could opt for the easier and cheaper version, and therefore choose cohabitation. Furthermore, the transition from cohabitation to marriage could be delayed and even forgone as a result of unfavorable economic circumstances. The latter two instances would lead to a rise in the share of cohabitation among all persons in a union.

Within the RWA framework, a basic change in the readiness condition, as described above, would not be sufficient. Concomitant changes in the other two conditions are equally necessary. In the Latin American context, we would therefore expect to identify major cultural changes as well, particularly related to ethics and morality, thereby lifting the stigma on certain forms of behavior, including cohabitation. Most likely, such changes are accompanied by further secularization and by changes in attitudes toward gender relations.

We address the readiness and willingness conditions in the next two sections. Discussion of the ability condition, which would require a detailed study of legal provisions and changes affecting the status of consensual unions, is beyond the scope of this chapter. Suffice it to say that national differences in trends related to cohabitation can also be the result of differences or shifts in such legal and institutional factors (cf. Vassallo 2011).

5.1 Cohabitation as a Response to Economic Shocks

Latin America has been characterized by both widespread social and economic inequalities and turbulent macroeconomic performance. After a period of dictatorships, a number of Latin American countries “re-democratized”, but policies aimed at diminishing the large differentials in standards of living resulted in inflation and outbursts of hyperinflation (Bittencourt 2012) Attempts at income redistribution during this populist phase were conducted through unfunded public deficits, which led to massive inflation, and ultimately to even greater inequality as the poor were affected more than the rich. In such instances the benefits of economic development realized before 1980 were often lost.

The timing, duration and severity of the periods of hyperinflation varied considerably from country to country. Roughly speaking, we can identify two patterns. The first was characterized by a very long period of inflation, but at peak annual levels during the 1980s that were generally below 30 %. The second pattern is a short period of inflation of such high intensity that money became worthless overnight. Peak levels of 1000 % inflation in a given year were common (Singh et al. 2005; Adsera and Menendez 2011). Obviously, the effect of such inflation spikes is felt for many years, and in the Latin American case, well into the 1990s. Examples of long duration inflation are Chile (already starting during the Allende presidency) and Colombia (Singh et al. 2005: 4). Examples of virulent hyperinflation are Brazil (2950 % in1990), Argentina (3080 % in 1989), Peru (7490 % in 1990) and Bolivia (11750 % in 1985). Such figures provide ample reason to advance the thesis that economic conditions could have been primary causes of the rise of the share of cohabitation in Latin America.

We mention three caveats regarding this explanation, however. As argued by Fussell and Palloni (2004) ages at first union remained remarkably stable throughout the second half of the twenthieth century and show a surprisingly low elasticity to such economic disturbances. The authors assert that economic conditions accelerated the fertility decline, but that, “as it has been for many centuries, the marriage and kinship system in Latin America continues to provide a system of nonmonetary exchange that parallels rather than competes with market systems.” (p.1211). In their opinion, the nuptiality system would provide a buffer against economic hardship, for both elites and the bulk of the population. But their research focuses on the stable ages at first union, not on the shift from marriage to cohabitation. Viewed from the latter perspective, much more “internal” change took place within the nuptiality system, and it remains possible that the more turbulent 1980s and early 1990s are at least partially responsible for accelerating the shift from marriage to cohabitation.

Our second caveat concerns the timing of both features, inflation and the rise of cohabitation. In two of the countries considered here, Brazil and Colombia, the largest increase in percentages cohabiting occurred during the 1970s, well before the shocks of the 1980s. During that decade, these percentages cohabiting continued to grow, but in two different inflation regimes. The Brazilian hyperinflation peak of almost 3000 % occurred in 1990, by which time the cohabitation share for women 25–29 had nearly tripled from some 8 % to 22 % (see Table 2.1). In Colombia, the 1980s inflation peak was much lower, at 33 %, and also long-term inflation was low by LatinAmerican standards – 16 % per annum for the second half of the twentieth century (Adsera and Menendez 2011: 40). Yet Colombia experienced the most pronounced increase in cohabitation, from around 20 % in 1970 to almost 50 % before the 1990 inflation maximum.

The two countries with the largest increments in cohabitation in the 1980s are Argentina and Puerto Rico. The former saw a hyperinflation peak of over 3000 % in 1989 and average annual inflation rates for the 50 years prior to 2003 of 184 % (ibidem). Puerto Rico, by contrast, experienced nothing comparable to Argentinean inflation levels, yet still recorded a noticeable rise in cohabitation before 1990. The Chilean example is also worth noting. Chile had an early hyperinflation peak of about 500 % during the 1970s, and again a more modest rise in the 1980s. Yet, Chile does not have the steepest rise in cohabitation by the year 2000. Similarly, also Mexico had its take off phase of cohabitation during the 1990s, and not a decade earlier when it had its high inflation regime.

The conclusion from these comparisons is the absence of a clear correlation between the timing and rise in cohabitation on the one hand, and the timing of inflation peaks or the overall rate of inflation on the other. Admittedly, a more precise time-series analysis is not possible since annual cohabitation rates, unlike marriage rates, cannot be computed. The entry into a consensual union is by definition an unrecorded event. The most one can say is that inflation and hyperinflation may have been general catalysts that strengthened the trend in the shift from marriage to cohabitation, but other causes must have been present as well.

Our third caveat points even more strongly in that direction. During the first decade of the twenty-first century, inflation rates in Latin American countries have fallen to much lower levels than during the 1980–1995 era, and yet, the upward trend in cohabitation has not abated. In fact, as the results for the 2010 census round indicate, the opposite holds to a striking degree in Uruguay, Argentina, Ecuador, Costa Rica and Mexico where a high rate of increase in cohabitation has been maintained (Table 2.1). Even Panama, which had the highest incidence of cohabitation throughout the entire study period, witnessed a further increase in cohabitation during the first decade of the new Century. Hence, it is now very clear from the 2010 census round that the rise in cohabitation is a fundamental systemic alteration and not merely a reaction to economic shocks.

5.2 Lifting the Stigma: Cohabitation and Ideational Change

As the RWA-framework posits, the switch to larger shares of cohabitation in all strata of the population would not have occurred had a major stigma against cohabitation persisted. Hence, the “willingness” condition must have changed in the direction of greater tolerance. Responses to the World Values Surveys indeed suggest the occurrence of a major change in crucial features of the ideational domain. We now turn to that evidence.

The European (EVS) and World Values Studies (WVS) have a long tradition often going back to the 1980s to measure major ethical, religious, social and political dimensions of the cultural system. Most Latin American countries have only one wave of the WVS, and a single cross-section is of course inadequate for our purposes. Moreover, unlike the EVS, the WVS-surveys measure current cohabitation only (“living as married”) but fails to catch the “ever cohabited” state, thereby confounding married persons with and without cohabitation experience.Footnote 2

For three Latin American countries with large shares of post-1960s “new” cohabitation we can at least follow the trend over time with an interval of 15 years. Argentina and Brazil had WVS waves in 1991 and 2006, and Chile in 1990 and 2006, with a subset of questions being repeated across the two surveys. Several of these questions are of particular relevance for our purposes since they shed light on the changes occurring in the various age groups in values pertaining to ethics, secularization and gender relations.

In Table 2.4 we have brought together the WVS results for the 1990–1991 and 2006 waves with respect to five ethical issues. For three broad age groups and both sexes we have measured the percentages that consider as inadmissible (“never justified”) the following actions: divorce, abortion, homosexuality, euthanasia and suicide. With the exception of abortion in Argentina and Brazil, there are major changes in the direction of greater tolerance, and in many, there is just about a landslide with reductions in the percentages “never justified” of 10 to over 50 percentage points. Furthermore, these changes are often just as large among the older men and women (50+) as among the younger ones.

Table 2.4 Attitudinal changes in ethical issues in three Latin American countries, by age and sex, 1990–2006
Full size table

By far the largest change noted in all three countries is the increase in tolerance toward homosexuality. The percentages who consider this as “never justified” are halved or, as in Chile, have been reduced to a third or even a quarter of their 1990 levels. In addition, a similar landslide can also be noted with respect to euthanasia. It equally occurs in the three countries, among both sexes and in all age groups. The change is again most pronounced in Chile. The reductions in percentages rejecting suicide and divorce are more modest compared to the massive change in the previous two items, but still very substantial and found in all age groups. And, as noted above, only the attitudes toward abortion show a mixed picture, with greater tolerance emerging in Chile, but not in Brazil and Argentina.

The latter exception notwithstanding, the data in Table 2.4 clearly indicate that a massive attitude change has taken place during the last two decades in favor of greater tolerance to forms of behavior or interventions that were largely tabooed before. This is obviously a cultural change which is entirely in line with what the theory of the “Second demographic transition” predicted (Lesthaeghe and Surkyn 2004; Lesthaeghe 2010).

The next set of items deals with secularization. The results for three sub-dimensions are given in Table 2.5: church attendance, roles of the church, and individual prayer. In all instances we measured the percentages who are at the secular end of the spectrum (no attendance, no prayer, church gives no answers). The results for the four items in Table 2.5 are very clear in the Chilean case: secularization has advanced to a remarkable degree and the trend is entirely in line with those described for the ethical issues in Table 2.5. The evidence for Argentina is more attenuated. There is a major increase in non-attendance, but a much more modest increase in doubts about the church being capable of addressing family issues and in men reporting no moments of private prayer or mediation. By contrast the church’s capacity to address social problems seems not to have suffered in Argentina.

Table 2.5 Attitudinal changes regarding religion and secularization in three Latin American countries, by age and sex, 1990-2006
Full size table

The Brazilian outcome differs substantially from the previous two countries: the landslide toward greater ethical tolerance is not matched by advancing secularization. Compared to the 1990 WVS-round, the 2006 one indicates falling percentages of persons never or very rarely attending church and falling percentages of persons doubting the role of the church. In fact, there is a clear rise in the proportions thinking that the church has a role to play in family matters. Only the percentages without moments of prayer and meditation have not changed in any significant direction. Overall, the Brazilian lack of secularization is not in line with international trends.

The results for four classic attitudinal items regarding family and gender are reported in Table 2.6. The Chilean results are again the most striking and totally in line with the expected trend: a sharp increase for men and women of all ages who consider marriage an outdated institution, a parallel decrease of respondents considering that a child needs both a father and mother, a marked increase of persons disagreeing with the statement that being a housewife is just as fulfilling (even among men), and a clear drop in the percentages stating that men should have priority when jobs are scarce. It should also be noted that the “feminist” shift is as pronounced among men as among women.

Table 2.6 Attitudinal changes in issues regarding family and gender in three Latin American countries, by age and sex, 1990-2006
Full size table

The Argentinean results again follow the Chilean pattern, but with more moderation. The increase in the percentages considering marriage an outdated institution is just as large, but the Argentinean public is still more convinced that a child needs both a father and mother. There are also mixed signals regarding gender equality: there is the expected increase in persons who disagree with the role of housewife being just as fulfilling, but there is no convincing decline in the opinion that men should have priority when jobs are scarce.

The Brazilian results with respect to the two family items are equally mixed, but different: there is no increase in the percentages considering marriage as an outdated institution, and even a drop among female respondents, but there is a systematic reduction in percentages considering that a child needs a complete parental family. The trend with respect to the gender items is more consistent: there is a rise in percentages disagreeing with the fulfilling nature of being a housewife and a clear drop in those giving men priority if jobs are scarce.

The question of “what flew under the radar” can now be answered partially. The ethical dimension has indeed undergone very large shifts during the period under consideration. This lends strong support to the thesis that tolerance for various sorts of non-conformist behavior, including the rise of “new” cohabitation in Chile, Argentina and Brazil, has increased quite dramatically, and that as a consequence, the W or “willingness”-condition in the RWA-framework has ceased to be a limiting or bottleneck condition. Obviously other changes that remain undocumented here could have equally contributed in creating more favorable R and A conditions for the Latin American cohabitation boom, but at least it is becoming clear that a cultural shift component is again a necessary (but probably not a sufficient) ingredient of a more complete explanation.

6 The Family Context of Cohabitation and Single Motherhood

Not only has there been a rise in unmarried cohabitation, but also in the proportion of single mothers (e.g. Arias and Palloni 1996; Castro-Martín and Puga 2008; Castro-Martín et al. 2011). Since these features are often linked to increased chances of poverty it is essential to know whether cohabitors and single mothers are living is nuclear households with presumably essentially neolocal residence or, by contrast are co-residing with parents (often three generation households) and/or other kin or unrelated persons in extended or composite households. In addition, a nuclear family context would be more in line with the notion of a “second demographic transition”.

In what follows we shall present the most important trends for the period up to 2000, since the reworking of the IPUMS individual pointers in the household composition files (Sobek and Kennedy 2009) into a new typology (see Esteve et al. 2012) for the 2010 census round has not yet been completed. But results can be presented for 13 Latin American countries. Also, we shall refrain here from giving further technical details, as these can be found in Esteve et al. 2012.

More often than not, the shifts in living arrangements of young women are considered without further reference to the possible presence of other kin or other non-relatives. This is not a major issue in situations dealing with European populations or populations with European traditions since the neolocal nuclear household is by far the dominant one. But matters change considerably when other populations are analyzed. In these instances the incidence of extended or composite household structures becomes of interest, not only in its own right, but also because such family or household structures can absorb or soften the effects of economic shocks, or alleviate the consequence of more precarious situations. In the first instance marriage or cohabitation without leaving the parental household could have been a response to the period of high economic instability and hyperinflation of the 1980s. In the second case single mothers could benefit both financially and in kind from the presence of parents, other kin, or even non-relatives. In what follows we shall analyze our Latin American version of the LIPRO typology (Esteve et al. 2012: 700–703) as to reveal to what extent the shifts documented in the previous sections occurred within the context of nuclear versus extended or composite households. To this end standard tables are extracted from the LIPRO-master table for women 25–29 which all have the same structure in studying, both per country and over time, the internal distribution of 5 individual positions over 3 household situations. The 5 individual positions are: single mother, cohabiting or married without children, cohabiting or married with children. The 3 household situations are: nuclear, extended with parents and possibly other kin or non-kin, and all other forms of extensions or composite structures without own parents. Here, we shall consider the prevalence of any form of extension (i.e. with parents, kin or non-relatives) for each of the 5 union subcategories. These percentages extended (or composite with non-kin) of all types are given in Table 2.7. The complement of these percentages gives the incidence of living in nuclear households.

Table 2.7 Percentage of women 25–29 living in extended/composite households by type of union, Latin American Countries, latest available census data
Full size table

Table 2.7 illustrates that very considerable proportions of young women 25–29 still live in extended or composite families. This is particularly so for single mothers, with figures typically ranging between two thirds and four fifths. Only in Bolivia and Puerto Rico are these proportions below 60 %.Footnote 3 The degree of splitting off from the parental or otherwise extended household upon the formation of a partnership, either through marriage or cohabitation, can be assessed in the next two columns: still a third to over one half of young childless women in a partnership are commonly found in extended or composite households. Only in Argentina and Brazil do we find lower figures of the order of one quarter. Equally remarkable is that the differences between the cohabiting and the married women without children in the percentages living in extended households vary substantially between countries, but with the percentages for childless cohabitors systematically being higher than for their married counterparts. This may indicate that further splitting off from the parental household occurs when a cohabiting union is converted into a married one. Regardless of the actual process, all of this means that cohabiting partners are accepted as residents in extended households in very much the same way as married spouses. In other words, cohabitation does not lead to more nuclear households being formed, and in countries with a strong tradition for coresidence of young couples with parents and/or others, this tradition is maintained for cohabitors as well.

As indicated, the incidence of co-residence varies substantially from country to country. In Argentina and Brazil, co-residence in an extended household is least common for cohabiting childless couples, and it is equally rare for childless married ones in these two countries and in Puerto Rico. At the high end of the distribution for both types of couples are Ecuador, Venezuela, Peru, Bolivia and Cuba, with percentages in extended households typically in excess of 40 %. As expected, co-residence with parents or other adults drops further for cohabiting and married women with children. There is still a slight tendency for cohabiting mothers to be found more frequently in extended households than for married mothers, but this tendency is not universal. More striking is the lasting difference between countries. Puerto Rico, Costa Rica and Brazil have fewer than 20 % of young married or cohabiting mothers living in extended households, whereas the figures for Venezuela, Peru, Bolivia, Panama and Cuba are still in range of 30–40 %.

There are two substantive conclusions to be drawn from these findings. First, the more precise nature of the “robustness” of Latin American families to the economic shocks of hyperinflation in the 1980s, as perceived by Fussell and Palloni (2004), lies in the fact that co-residence with parents or others remains the rule for single mothers, and also remains very common for both cohabiting and married couples without children. And second, there is a caveat with respect to the Latin American convergence to the pattern of the “Second demographic transition” (SDT). The sheer size of the cohabitation boom and the de-stigmatization of unmarried unions definitely fit the SDT prediction, but the convergence to a purely western pattern is only a partial one given that significant proportions of childless cohabiting couples and a still noticeable percentage of cohabiting parents are not living in a nuclear household but in extended and/or composite ones. For such couples it is harder to imagine that cohabitation would be merely a “trial marriage” between two individuals. Hence for several countries there is a clearly distinct Latin American version of one of the key aspects of the SDT, and it is produced by the historical context of continued robustness of co-residence in extended households for a significant segment of the population. For the others, however, and they are a majority (9 countries of the 13 considered here), cohabitants do live predominantly in a neolocal and nuclear setting, and for them the convergence to the western SDT pattern is much more likely.

7 Conclusions

The reconstruction of the share of cohabitation in the process of union formation of both men and women in 665 Latin American regions indicates that there has been a real “cohabitation boom” taking place since the 1960s in some instances and accelerating during the 1990s in most. This holds particularly, but not exclusively, in areas which had relatively low levels of “old” or traditional cohabitation with a historical ethnic background. Furthermore, the upward trend shows no signs of abating during the first decade of the twenty-first century, and latecomers such as Mexico and Uruguay have caught up with the others. Hence, the lion’s share of the boom is due to “new” cohabitation. Moreover, the negative gradient of cohabitation with female education is somewhat alleviated over time since the rise in cohabitation affected all educational categories, with the middle educational groups and the more educated catching up to a significant extent.

This raises the question whether or not this feature signals a partial convergence of Latin American countries to the European pattern of the so called “second demographic transition”. The discussion of this question has already emerged in the Latin American literature (García and Rojas 2004; Cabella et al. 2004; Rodríguez Vignoli 2005; Quilodrán 2008; Castro-Martín et al. 2011; Salinas and Potter 2011; Covre-Sussai and Matthijs 2010). Two arguments are offered here in favor of such a convergence. Firstly, on the basis of both the negative cross-sectional gradient with education and the steep rises in female education, one would expect the share of marriage to gain importance, and not the share of cohabitation. Secondly, for three major countries with a sizeable increase in “new” cohabitation (Chile, Brazil, Argentina) data from two rounds of the World Values Studies show major changes, if not a landslide, in the direction of greater tolerance for previously tabooed behavior or actions, such as euthanasia, homosexuality, and suicide. Moreover, several other attitudes in favor of greater secularism, of non-conformist family arrangements, or more egalitarian gender relations emerged during the 15 year period documented by the WVS. These ideational changes, and particularly those in ethics, are indicative of the fact that the cohabitation boom has indeed developed in a context of growing individual autonomy and greater overall tolerance.

The expansion of cohabitation and of parenthood among cohabitants, or the “non-conformist transition”, is not the only hallmark of the SDT. The other major ingredient is the so called “postponement transition” with the shift to older ages of both nuptiality and fertility. In Western and Northern Europe, both the non-conformist and the postponement parts occurred more or less simultaneously. In advanced Asian industrial societies, the marriage and fertility postponement preceded the hitherto modest increase in cohabitation by three decades. A similar timing gap was witnessed in Southern Europe. The Latin American experience provides an illustration of the reverse, with the “non-conformist transition” preceding the postponement one. If that proposition holds, we should now be looking out for rises in ages at first birth and further drops in fertility to below replacement levels.