Marital status in a developing economy: gender, ethnicity and family structure

Abstract

This article employs population survey data for Trinidad and Tobago to investigate the likelihood of being married, both legal and common law in the present or the past, as compared to being never married, in a developing economy. The article finds support for the sex-specialized theory of marital formation, with men valued more and women valued less for their economic contributions to the family. Socioeconomic status and ethnicity play an important role in marital formation with sex specialization less likely to occur among Africans than the more advantaged ethnic groups. One of the consistent findings is that family structure matters crucially in predicting both legal and common law marriages. For example, living in a large, multigenerational household tends to impact the likelihood of marriage, as compared to remaining single, in opposite ways for men and women, with men in large households more likely to be married but women less likely to be married.

Notes

¹ Common law marriages require a legal basis for the marriage, usually by a cohabitating couple living together long enough to meet the legal requirements for the marriage to be considered legally valid. In general, cohabitation refers to both common law marriages, where allowed, and a couple living together without the benefit of either a more formal marriage or a common law marriage. This data used in this article generally refers only to common law marriages. The term legal marriage will be used to refer to the more formal type of marriage.

² Individuals with Indian ancestry were originally imported into the country from India as workers after the end of slavery. There are virtually no surviving natives in Trinidad and Tobago. In fact, the CSSP survey does not include native as a possible answer when soliciting a person's ethnicity.

³ See, for example, Lichter et al ., 1991; Bumpass et al ., 1991; Mare and Winship 1991; Lichter et al ., 1992; Manning and Smock 1995; Sanchez et al ., 1998; Moffitt et al ., 1998.

⁴ That is, Xβ _i = ln(P_i /P_N ), where β _i is the vector of coefficient estimates for i = married, commonlaw, once married, once commonlaw, X is the matrix containing observations of the explanatory variables, P_i is the respective probabilities and P _N is the probability of being never married. See Green (1993) for more details.

⁵ Green (1993) notes that the standard errors for the marginal effects can be estimated but are complex and recommends using those for the coefficient estimates.

⁶ See Winkler (1997) for evidence on how cohabitors in the US, including common law spouses, treat other sources of household income including their partner's income.

⁷ Full results for these models are available from the author upon request.

⁸ That households are primarily organized around, at most, a single married couple is evidenced by the fact that only 2.3% of all households have either multiple married couples, multiple commonlaw couples, or both present.