There is evidence of an increase of union instability in Brazil and at the individual level women with more than one union are usually reported to have higher fertility. At the same time there has been an increase of consensual unions usually known for having a higher rate of dissolution than formal ones. The Brazilian total fertility rate fell from 6.5 in 1930 to 2.1 in 2000. The first evidence showed that the level of fertility is higher for those women with more than one union in any age group. Despite this, at a macro-level it is clear that Brazil showed a decline of fertility during a period of increase of union dissolution. The current literature stresses the attention on the role of sterilisation and abortion on dramatic fertility decline. However there are no recent studies that analyse the complex relationship between union dynamics and childbearing mainly due to the lack of data. Brazilian couples report serious lack of communication on reproductive matters and in particular for women of low-income status, reproductive choices are often dictated by the need to keep their partner. It is for this reason that to get a better understanding of the current fertility trends, there is a need to understand in depth the mechanisms that interact in union dynamics and reproductive decision-making.

The aim of this thesis is to analyse the union formation and dissolution behaviour and their interaction with fertility in Brazil. The second aim of this thesis is to utilise the Demographic and Health Surveys (DHS) calendar section to analyse union and reproductive histories. The DHS calendar, generally used for studies on contraceptive use, allows the simultaneous analysis of parallel event histories of union status changes and fertility, an analysis that has not previously been done for Brazil. The first part of the analysis is dedicated to the comparison of the modelling of the history of union status change as a single equation with the multiprocess modelling of union formation, dissolution and conception outcome using Lillard’s model (1993). This type of analysis allows to study the relationships between union dynamics and their effect on fertility. Using the 1996 DHS it is demonstrated that union formation and union dissolution have a positive effect on the level of fertility. Furthermore it is shown that, in particular for women in their second or higher order union, childbearing could be used as a mean to strengthen their relationship.

Analysing in depth the relationship between the timing of sterilisation and union dynamics using discrete time modelling it is demonstrated that, women with a history of unstable relationships delay the timing of sterilisation. This could prove that the relationship between union instability and fertility is only one aspect of the overall relationship between union dissolution and reproductive matters.

The second part of the analysis is dedicated to the macro analysis of the actual effect of union instability on the level of fertility. The effects of three main components in the level of Brazilian fertility that have shown interesting trends in the last two decades have been decomposed: adolescent fertility, union instability and sterilisation.

In the absence of other methods, the fertility decline in Brazil is primarily due to the increased prevalence of female sterilisation. It is shown that the decline, powered by female sterilisation, would have been more rapid in the absence of the increase in teenage pregnancy and union instability.