Click here for search results

Newsletter

Site Tools

Infant Mortality over the Business Cycle in the Developing World

Research Brief narrow banner 
Nov 9, 2007, Sarah Baird, Jed FriedmanNorbert Schady

Research on income fluctuations in per capita GDP and child survival in 59 countries shows a strong, negative association between changes in per capita GDP and infant mortality.  The results suggest that over 1 million excess deaths have occurred in the developing world during 1980-2004 in countries experiencing economic contractions of 10 percent or greater.

 

 

 

The relationship between aggregate income and infant mortality in a large sample of developing countries is explored in a new study.1 The study analyzes child (rather than adult) health to limit the potential for reverse causality from health to income. It also accounts for possible fluctuations in GDP from external shocks such as civil war and severe weather in order to further isolate the potential causal impact of income changes on health.

Infant survival is sensitive to fluctuations in aggregate income

The data on births and deaths are based on 123 Demographic and Health Surveys (DHS) covering 59 countries in three regions: Africa (33 countries, 68 surveys), Asia (14 countries, 27 surveys), and Latin America (12 countries, 31 surveys). The earliest surveys were carried out in 1986, the latest ones in 2004. The assembled information represents 760,000 women and 1.7 million births.

The estimation strategy controls for the country-specific deterministic trends or non-stationary processes in the observed GDP and infant mortality series through de-trending or first-differencing, and then regresses single-year period innovations or deviations in infant mortality on GDP.

Table 1 shows the significant association between the stationary or de-trended components of per capita GDP and the infant mortality rate: a 1-percent contraction (or negative deviation from trend) in per capita GDP is associated with an increase in infant mortality of between 0.18 and 0.44 deaths per thousand children born.

Table 1.  Change or detrended infant mortality rate on change or detrended log per capita GDP, by various methods of trend accountingTable 1 for Infant Mortality over the Business Cycle

 

 

 

 

 

The analysis shows that the changing composition of women giving birth cannot account for the bulk of the association between infant mortality and GDP

In order to explore the idea that the composition of women giving birth may change with economic circumstances, and that this may have an effect on mortality, the underlying household data was used to assess changes in infant mortality as a result of changes in the composition of women giving birth or, rather, changes in the probability that a child born to a woman of given characteristics dies.

Table 2 controls for characteristics highly correlated with the probability of child survival such as the place of residence at the time of the survey, infant birth order, infant gender, whether or not the child was part of a multiple birth, and mother’s years of education and age at the time of the birth. The second row reproduces the results from Table 1, which do not control for mother characteristics, for ease of comparison.

Table 2. Change or detrended infant mortality rate on change or detrended log per capita GDP, controlling for birth and mother characteristics Table 2 for Infant Mortality over the Business Cycle

 

 

 

 

 

 

 

 

 

The comparison in the table suggests that the coefficient on per capita GDP generally falls by a relatively small amount when adjustments are made for the changing composition of women giving birth. For example, in the first-differenced specification, the coefficient in the first row of Table 2 is approximately 20 percent smaller than the comparable coefficient in Table 1. These data suggest that the changing composition of women may play some role, but cannot account for the bulk of the association between infant mortality and GDP we observe in our data.

back-to-top image

The causal relationship between aggregate income and infant mortality becomes clearer when controlling for possibly relevant third factors

The main concern with causal inference is less likely to be reverse causality from infant health to economic circumstances—an important question when interpreting associations between adult health status and income—and more likely to be omitted variables that affect both aggregate economic conditions and infant mortality.

Two likely candidates for omitted variables are conflict, including civil war, and weather shocks, including droughts and floods. Conflict may directly result in infant deaths, or disrupt the provision of health services. Large-scale droughts or floods may change the health environment and the transmission of communicable disease. Further, conflicts, floods and droughts may all result in economic contractions. This raises the possibility that the association we estimate is spuriously driven by these third factors.

To address these omitted variables all country-year observations that included conflict or a weather shock were removed from the sample. The overall results are unchanged suggesting these third factors are unimportant.

back-to-top image

The association between changes in infant mortality and changes in GDP is stronger when GDP shocks are negative and large

Figure 1 imageFigure 1 shows the relationship between log per capita GDP and infant mortality by the magnitude and sign of the changes in GDP. It presents a non-parametric regression of infant mortality rate change on GDP change, both unadjusted and controlling for mother's characteristics.

The evidence that large negative shocks to GDP have the largest effects on infant mortality is consistent with an inability of households or governments (or both) to smooth consumption over the business cycle. Credit constrained households may be unable to keep up investments in child nutrition or health care, especially when the shock to GDP is large. A simple calculation suggests that 1,025,000 excess deaths have occurred in the developing world during 1980-2004 in countries experiencing economic contractions of 10 percent or greater.

Implications for policy

The variability in the relationship between macroeconomic crises and child health, as evidenced by the relatively large standard errors in Table 1, may reflect the results of different policy responses. Cursory country-specific comparisons suggest that household, government, and donor responses to macroeconomic collapse can mitigate negative impacts on child health.


Note

1. This brief is based on "Infant Mortality over the Business Cycle in the Developing World" by Sarah Baird, Jed Friedman, Norbert Schady, World Bank Policy Research Working Paper 4346, September 2007.

Researchers

SARAH BAIRD is a Postgraduate Scholar at the Center on Pacific Economies at the Graduate School of International Relations and Pacific Studies at the University of California, San Diego. Her research interests include health, risk, and program evaluation. She is currently working on a conditional cash transfer experiment in Malawi that looks at the impact of schooling on HIV risk. Email: sjbaird@ucsd.edu

JED FRIEDMAN is an economist in the Development Research Group (Poverty Team). His research interests include the measurement of poverty dynamics and the interactions between poverty and health. He is currently involved in assessing the socioeconomic impact of the 2004 Indian Ocean Tsunami as well as the efficacy of aid programs implemented in Sumatra, Indonesia, in response to the disaster. He is also investigating the linkages between micronutritional status, general health status, and individual labor productivity and other work outcomes. Email: jfriedman@worldbank.org

NORBERT SCHADY is a Senior Economist in the Development Research Group (Human Development and Public Services Team). His research interests include education, labor markets, political economy, and the impact of macroeconomic shocks on human capital outcomes. Email nschady@worldbank.org

bck-to-top image




Permanent URL for this page: http://go.worldbank.org/RZO6EBYQQ0