ISER Working Paper Series 2006-30
State dependence and causal feedback of poverty and fertility in Ethiopia
01 Jul 2006
Whereas many low-income countries has experienced substantial decline in TFR together with economic growth, Ethiopia is one of the countries where fertility levels have remained high, together with consistently high poverty levels. The contributing factors are of course many, including poor employment prospects, primitive production technology in rural areas, lack of family planning, low human capital and poor education infra-structure, and last but not least, strong social norms associated with women's role in society. Whereas much has been said about the associations between poverty and fertility, existing data sources have prevented analysis of a more causal nature. Using longitudinal data set from three comparable waves from rural and urban Ethiopia together with random effect models, our paper goes a little further in terms of establishing causal mechanisms.
An important finding of our paper concerns state dependence in poverty. This is particularly strong in urban Ethiopia, and suggests that the experience of poverty leads to further future poverty. There is also evidence of state dependence of poverty in rural Ethiopia, but in this case the estimates are more sensitive to the equivalence scale. There is also state dependency in childbearing, in that past fertility events tend to influence further childbearing. The effects depend however on the age of children and there are important differences between urban and rural areas. Whereas recent childbearing events (i.e. presence of young children) in urban areas tend to discourage further childbearing, it encourages more children in rural areas. However, we find very little evidence of any causal feedback mechanism from poverty onto fertility, independent of the sample used. Moreover, the feedback from childbearing onto poverty is also rather weak once state dependence is controlled for.
Another important finding of our study is the significance of child labour as a symptom of both high fertility and high poverty. Whereas we do not find any significant effects in urban areas, possibly due to small sample size, we do find that child labour is closely related to both fertility and poverty in rural areas. Reducing child labour enhances children's possibility of becoming non-poor adults in the future. Of course, rural Ethiopia are still facing endemic school drop out rates, especially among girls. The poor educational infra structure is likely to perpetuate the low enrolment rates. Our results provide strong support for the policy recommendations in SPRPR, which highlights the importance of increasing enrolment rates in rural Ethiopia.
We also find that household composition matters. In particular we find that the more adults present in the household, the higher is poverty. Moreover, extended household where there are more than two generations living in the household (e.g. presence of grand parents) are not always more likely to be poor, but tend to have higher fertility. On one hand large families might simply reflect traditional coping mechanisms where friends and relatives pool resources to deal with economic hardship. But our measure of extended family relations might also reflect traditional attitudes and social norms, in which men are perceived successful if they have many children, and women are expected to specialise in household production and rearing of children. In so far this is the case, presence of social norms is potentially important in explaining both high fertility and high poverty.
Our analysis also shows that education have strong explanatory power in urban Ethiopia, both for fertility and poverty. As expected, higher levels of education either of the household head or other household members, reduce the poverty incidence. The role of education is not equally strong in rural areas. This is perhaps not unexpected, since the distribution of education in rural areas is highly skewed in that very few have much formal education, a fact reflected by extremely high illiteracy rates. Moreover, since rural Ethiopia is dominated by primitive agriculture, returns to schooling might not be particularly high. Consequently, our study suggests that there is a great need to increase enrolment rates partly to offset the high incidence of child labour, which in turn is related to both poverty and fertility. At the same token, the government should ensure that the school drop out rate is reduced.
Whereas our study is able to detect causal mechanism underlying poverty and fertility, one should be aware of potential shortcomings. One concern relates to the longitudinal aspect in that our sample is limited to three waves only. It would be useful to assess the relationship between poverty and fertility with a longer time span. Second, the initial condition for childbearing is based on the household roster and not a complete fertility history. Thus, we do not capture children that might have left the household, nor do we capture the potential effect of infant mortality. We have also seen that some of the parameters are sensitive to the use of equivalence scales. Though the WHO equivalence scale is detailed and based on diet requirements, which seems sensible in a country like Ethiopia, we could extent the analysis by estimating equivalence scales by fitting Engel curves using data from the surveys themselves.