The relationship between sex ratios and marriage rates in South Africa

Abstract

We investigate the relationship between alternative definitions of sex ratios and marriage outcomes among African and white women in South Africa. In contrast to marriages among whites, African marriages in South Africa traditionally have involved the payment of bridewealth (or ilobolo) by a husband to the prospective wife's family. Using matched data from the 2001 Population Census and the South African Labour Force Surveys, we find that among Africans, both the quantity and quality of unmarried men relative to women in local marriage markets are significant predictors of marriage. However, economic-based measures of marriageability have a larger effect on marriage outcomes than simple sex ratios. These findings are consistent with the argument that bridewealth payments act as a financial constraint to marriage among African couples, raising the marriageability criteria of men. In contrast, we find mostly insignificant results for the relationship between sex ratios and marriage outcomes among white women.

Keywords:

• marriage
• sex ratios
• bridewealth
• South Africa

JEL Classification:

• J12
• J11

Acknowledgements

We thank two anonymous reviewers for the helpful comments received, and Economic Research Southern Africa for supporting this research.

Notes

¹ The term African is used to describe black South Africans who make up the majority of the population (approximately 80%).

² It is also possible that attitudes to marriage, particularly in the context of high rates of Human Immunodeficiency Virus (HIV) infection in South Africa, contribute to low and declining marital rates. However this view is challenged by data collected in a nationally representative attitudinal survey, the South African Social Attitudes Survey (SASAS) of 2005, where almost 90% of African unmarried adults (20 to 39 years) reported that they wanted to be married someday (own calculations, SASAS, 2005). In addition, couples continue to form other types of partnerships despite the risks involved. South African household survey data from 1995 and 2008 suggest that while marriage rates have been on the decline among Africans, rates of cohabitation have trebled from about 5% to 15%. Africans are also almost three times more likely than whites to cohabit with a partner, which would be consistent with the idea that for many African couples marriage may not be a financially viable option. Nonetheless, data from the SASAS (2005) also suggest that for the majority of unmarried African men and women (64% of those 20 to 39 years), cohabitation is not seen as an acceptable alternative to marriage, which would help explain why, in the face of low marriage rates, cohabitation rates among African have not risen by even more.

³ There is no information collected on the payment of ilobolo in national household surveys in South Africa. Information collected in the 1998 wave of a regionally-based panel study (the KwaZulu-Natal Income Dynamics Study) gives some indication of the extent to which ilobolo is still practiced and its value. Of the married respondents aged 60 years or younger, three quarters reported making ilobolo payments with marriage. The average value of ilobolo reported for people married from 1985 to 1998 was approximately 20 000 Rands (2000 prices) or almost 13 times the average monthly real earnings of African men in the 1998 sample (Casale and Posel, 2010). This value is consistent with reports in other literature of ilobolo typically ranging from 10 000 Rands to 25 000 Rands (Kaarsholm, 2005; Gustafsson and Worku, 2006).

⁴ Hosegood et al. (2009, p. 284) write: ‘The legacy of the early Natal administrators is that they co-opted and codified bridewealth. While historically the amount of bridewealth was negotiated by the families involved and was rarely paid in full before the marriage took place, the Natal code subjected Zulu women to a fixed and very high bridewealth of eleven head of cattle or their equivalent value’.

⁵ The LFS from March 2000 to March 2004 do not distinguish between marriage and cohabitation. Consequently our measures will overstate the true marriage rate, and because cohabitation rates are typically higher among African women than among white women (Budlender et al., 2004), the difference between white and African marriage rates will be underestimated.

⁶ It is possible that men who are in polygamous marriages will be considered part of the potential set of partners available to unmarried women. However, a negligible percentage of African men (0.14%) in the age group of 20 to 39 years report being involved in polygamous marriages in the 2001 Population Census. We therefore do not adjust the sex ratio to reflect the potential availability of these men.

⁷ Because of a skewed earnings distribution in South Africa, mean earnings are considerably higher than median earnings, and given the legacy of racial segregation in the labour market, mean and median earnings for young white men are significantly greater than those for young African men.

⁸ Among whites, the age-specific district sex ratio in 2001 ranged from 0.507 to 3.468 and among Africans, from 0.249 to 2.471. Relatively low sex ratios among Africans compared to whites are partly explained by a higher proportion of the male population being incarcerated and by low sex ratios in districts with larger populations.

⁹ This resulted in the loss of five out of 7705 observations in the sample of young African women aged 20 to 30 years and 70 out of 635 observations in the sample of young white women.

¹⁰ We applied an extension of the Blinder–Oaxaca decomposition technique to binary outcome variables, developed by Fairlie (2006), to decompose the racial gap in marriage rates.

¹¹ The next lowest marriage rates by province were in the Eastern Cape and North West provinces, where approximately 19% of African women in the 20 to 30 age cohort were married.

¹² A migrant worker is defined as an individual who is reported as a member of the household but who is not resident in the household and is away for at least a month each year to work or to look for work.

¹³ This index is calculated using the LFS 2003:02 data as there is no information on migration in the 2001 Population Census (or in the 2001 LFS). The LFSs which include a migration module do not collect information on the age of the migrant worker, and we therefore could not identify a ratio of young migrants specifically.

¹⁴ We also tested whether our results remained robust for a reduced age cohort of women aged 25 to 30 years; and when we restrict economic-based measures of marriageable males to men with formal (nonfarming) employment. Formal sector employment, which tends to be more secure than informal sector employment, may offer a better indication of a man's future economic status and therefore his marriageability. We find that the pattern and significance of results remain unchanged for both Africans and whites in both sets of estimations.