Published on Development Impact

What happens when the power goes out? Using blackouts to help understand the determinants of infant health

This page in:

Low birth weight, usually defined as less than 2500 grams at birth, is an important determinant of infant mortality. It is also significantly associated with adverse outcomes well into adulthood such as reduced school attainment and lower earnings. Maternal nutrition is a key determinant of low birth weight and it’s no surprise that nutrition interventions targeted at pregnant mothers can have significant impacts. Maternal nutrition and general maternal well-being are, in turn, partly a function of household resources. Income support programs such as conditional cash transfers can also help to raise infant birth weight. The link that runs from income support during pregnancy through maternal nutrition and reduced low birth weight to increased schooling attainment and adult productivity highlights the potentially life-long impacts of such temporary transfers.

But what about the down-side risk of temporary income fluctuations - do short-lived negative income shocks have equally significant effects on low birth weight? Households may be able to prioritize the consumption and care of pregnant mothers during adverse shocks, but of course households must know about the pregnancy in the first place. This knowledge doesn’t usually manifest until after the first 6-8 weeks of pregnancy and those initial weeks of pregnancy are also critical ones to ensure the health of the fetus. One recent study by Alfredo Burlando focuses on this critical window of time when households do not yet have sufficient knowledge and thus do not sufficiently protect against the changing economic circumstances.

In May of 2008, the undersea cable that brings power to the Tanzanian island of Zanzibar was ruptured, plunging the island into a blackout that lasted 4 weeks. As a result, households employed in sectors such as manufacturing or tourism that relied on electricity experienced income declines while households in more traditional sectors such as farming did not suffer noticeable shortfalls. Fortunately any income decline was short-lived – the power was only out for 4 weeks – and in a matter of months income in all affected sectors had recovered to previous levels. Despite the brief duration of this income shock, could there have been any long-lasting consequences?

Well it turns out that infants born 7 to 9 months after the blackout were significantly smaller – an average of 75 grams smaller – than infants born within 6 months of the start of blackout or beyond 9 months after its end. This reduction translates into an 11% increase in the probability of a low weight birth. Burlando proposes reduced nutritional intake and heightened maternal stress, brought on by the blackout induced income shock, as the main transmission mechanism for lower birth weights.

Because this is an observational study that utilizes a natural experiment of sorts, Burlando needs to consider, and rule out, alternative explanations for the observed pattern in birth weights. For example, if the food supply on the island was affected by the blackout then this may directly affect infant health. However the prices of key food items were stable over the period suggesting little to no disruption in food supply. Additionally, the blackout did not appear to affect the functioning of the health system and health services continued at a regular pace.

Another potential confounder is the selection of women, such as teenagers or those with less education, that are more likely to have low weight births into pregnancy as a result of the blackout. Although evidence was found of a change in fertility behavior due to the blackout, with 11% more births approximately nine months after the blackout – clearly something else happens when the power goes out – the bump in fertility was experienced by mothers of all characteristics in roughly equal proportions.

The findings suggest that women who were known to be pregnant at the time of the black out, i.e. those who were visibly pregnant, received insurance from the shock where as women who did not realize they were yet pregnant (or who had conceived during the blackout) did not receive the same protection.

For me, the take away messages from this study are threefold:

-          These findings highlight the importance of behavioral responses and that people in the face of a crisis can be resilient when they are armed with relevant knowledge – households with women who knew they were pregnant apparently prioritized maternal nutrition. It also underscores the obvious point that any protective program that targets pregnant women faces the challenge of improving the informational barriers that prevent early pregnancy awareness.


-          The study also highlights the long-lasting effects of even very brief income shocks if (a) they occur at critical moments in fetal development and (b) households cannot fully smooth consumption or otherwise insure themselves from temporary declines.


-          Finally, this study highlights the potential uses of administrative data in evaluative research. The results are estimated from 16,000 paper birth records taken from the main maternity hospital in Zanzibar and entered by the researcher into electronic form. As administrative data systems in a variety of sectors transition to electronic form, the types of questions that researchers can address – especially those that leverage unexpected and temporary shocks where surveys are impractical – will continue to surprise.



Jed Friedman

Lead Economist, Development Research Group, World Bank

Join the Conversation

The content of this field is kept private and will not be shown publicly
Remaining characters: 1000