Share

BREAD Working Paper No. 288, November, Subha Mani, John Hoddinott, and John Strauss (2010).

This article identifies path dependence in educational outcomes in rural Ethiopia, arguing that shocks to school enrolment significantly impact future attainment.

Households in developing countries frequently face the prospect of income shocks.  Those shocks can lead to adverse consequences, including pulling children out of school.  In this paper, Mani, Hoddinott, and Strauss investigate the long term impact of such shocks on education.  They ask, “the importance of these losses from a policy perspective depends partly on whether such shocks induce path dependence.  That is, do transitory shocks have permanent consequences?  Or put another way, is past history destiny?” (3)

Other scholars have conclusively identified path dependence in some areas, such as nutrition.  Birth year rainfall in Indonesia affects adult height, for example.  Path dependence has also been identified in child height in Zimbabwe. 

Could this type of path dependence extend to educational attainment?  The authors suggest yes.  They argue that “Schooling outcomes – such as the decision to continue or withdraw from school, or to enroll having previously not enrolled in school – would seem to be intimately linked to past schooling decisions which themselves were influenced by prior community, school and home resources” (3). 

Methodologically, the authors explore this question with data from three waves of the Ethiopian Rural Household Survey conducted in 1994, 1999, and 2004.  A longitudinal data set was constructed from this survey following primary school aged children (7-14 years old) through the three waves of the survey.  Household attrition was minimal.  Seasonality does not bias the dataset as the survey waves were conducted in roughly the same months.  Beyond the correlation between interrupted schooling and attainment, the authors attempt to establish causality with econometric instrumental variables methods.  They present various robustness checks that suggest that these methods and causal findings are valid.

Their results suggest that there is path dependence in educational outcomes in a rural, developing setting.  In particular, “a child who is enrolled in the last period is 32 percentage points more likely to be enrolled today compared to his counterpart who was not enrolled in the last period and… past levels of relative grade attainment affect current levels of this outcome” (4).  Path dependence was much stronger for girls.  A girl enrolled in the last period was 69 percentage points more likely to be enrolled today compared to her counterpart, while the equivalent boy was 21 percentage points more likely to be enrolled compared to his counterpart.  Path dependence was also stronger for wealthier households, with an 81 percentage point difference for children from high income households and a 7 percentage point difference for those from low income households.

Therefore, students’ schooling history can have a “strong” impact on their future schooling outcomes.  In summary, “Any lags and delays that affected progression in the past will have a permanent impact on final grades accumulated” (18).  These findings have significant policy implications.  Educational path dependence in a developing country context suggests that early schooling shocks should be cushioned in order to facilitate human capital development.