Father’s day post: what does paternal death tell us about the importance of fathers?

This is a guest post by Joachim De Weerdt, Kathleen Beegle and Stefan Dercon.

Last Sunday many (but not all) countries around the world celebrated father’s day, but what does the development literature say about the role of fathers?

Our thinking on this issue is still very much influenced by earlier papers that used variation in income to identify differences in spending patterns and bargaining power. Nearly three decades ago the seminal article of Duncan Thomas showed that in Brazil (unearned) income under the control of the mother had more desirable effects than similar income in the hands of the father. Five years later a paper by John Hoddinott and Laurence Haddad showed how, in Côte d'Ivoire, increasing women’s share of cash income increased the budget share of food and decreased the budget share on alcohol and cigarettes.

This is not just an issue in developing countries. An influential paper by Shelly Lundberg, Robert Pollak and Terence Wales looks at what happened to household expenditures when the UK started transferring child allowances to mothers instead of fathers in the late 1970s. This transfer of income ‘from the wallet to the purse’ increased expenditures on women’s and children’s clothing, and decreased expenditures on men’s clothing. More recently Alex Armand and his coauthors use an experiment from Macedonia that essentially randomizes the gender of the recipient and finds that cash to women gives more socially desirable outcomes than cash to men.

But not all studies find that giving cash to women is better. Najy Benhassine and coauthors find that in a cash transfer program in Morocco that aimed to increase school participation it did not matter whether the transfers were given to the father or the mother of the child. Richard Akresh, Damien de Walque and Harounan Kazianga find that cash to fathers improves children’s health in Burkina Faso whereas cash to moms did not. And Suresh de Mel, David McKenzie and Christopher Woodruff find that grants given to microenterprise owners generated larger profits for male owners compared to female owners. In a previous post David refers to that study to make the point that we should not let the focus on intra-household allocation detract us from the bigger picture of raising total household income. Cash not spent directly on children may not be wasted if it serves to raise future household income.

So much of the debate is centered income: who receives it, from which source it is derived and asymmetries of information regarding the size of the income. Markus gives some thoughts on where this research should go next.

While this research continues to grow, another strand of the literature is looking at a similar question from a different angle. Can we learn something about the importance of a parent by looking at situations in which the parent is no longer there? During the early nineties there was increased attention for the plight of orphans, in particular in regions that were heavily affected by HIV/AIDS. On the one hand we have studies that use a large number of cross-section data sets and have wide geographical coverage (e.g. here and here). On the other hand there are those with a more narrow geographical focus, but using panel data to get a better handle on causality.  Our own work, using the long-run Kagera Health and Development Survey (KHDS) panel data set from Tanzania, showed that maternal orphans suffer a permanent gap of 1 year of formal education and attain 2 cm lower height in adulthood. But we found no effects for paternal orphans.

Most orphan studies have tended to focus on schooling (and a few look at health) of children, but little is known on the psychosocial effects of orphanhood. During the latest data collection round in Kagera, Tanzania we included the 10-item Rosenberg Self-Esteem Scale (RSES). Self-esteem has been shown to be an important determinant of earnings and the RSES, even though it was developed in and for a Western context, has been shown to travel well across different contexts.

Our latest paper, published in JAIDS Journal of Acquired Immune Deficiency Syndromes, analyses the relation between orphanhood during childhood and self-esteem in adulthood. We start from a sample of 1,108 non-orphaned children who were interviewed in 1991-94 and re-interviewed in or approaching adulthood in 2010. Over one third of these children had lost one or both parents in between the interview waves. We compare self-esteem in adulthood between respondents who grew up as orphans to those who did not and differentiate the effect by the gender of the deceased parent. Obviously parental death is not a random event. We see for example that low socio-economic background predicts future paternal death. We therefore use the baseline data on these children to control for their observable, pre-orphanhood characteristics.

We find a reduction of 0.2 standard deviations in our self-esteem measure for paternal orphans, but no effect for maternal orphans. The effects become stronger when the death occurs during the child’s teenage years. So Dad’s may matter in ways we don’t capture with our more traditional set of outcomes.

Some concluding remarks:

1. First, a note on identification. One improvement of our paper over cross-sectional work is that we can control for pre-orphanhood characteristics from the baseline. In our work on health and education effects of orphanhood this turned out to matter a lot. In our latest paper on self-esteem, even though we show that parental death was related to observable baseline characteristics, not controlling for them actually made no difference for the end-results. Of course, we only know this because we collected them in the first place!
2. Two of our papers on orphans were published in medical journals in the hope that they will reach a wider audience. One in the International Journal of Epidemiology (impact factor 7.5) and one in the JAIDS Journal of Acquired Immune Deficiency Syndromes (impact factor 4.6). It takes some getting used to the different way of writing to what we are used to in economics journals.  Among other differences, there is a structured abstract and the articles are much shorter, to the point of being blunt.
3. While KHDS was set up in the early nineties to study the effects of HIV/AIDS the long-run nature of the data, the low attrition rates, the tracking of migrant respondents and the influx of refugees in the region have created a lot of demand for this data. We know of 112 papers written on the data set. Please let us know if your KHDS paper is not listed here.