Treading water: The long-term impact of the 1998 flood on nutrition in Bangladesh
Introduction
Bangladesh was struck by the worst flood in over a century in the summer of 1998. The flood began in July, and the water finally receded in September. The level of the water rose slowly throughout the summer, and at its peak it covered two-thirds of the country. The consequences for sanitation and health were worst at the onset of the flood and in the months of July and August.2
In addition to the direct impact of the flooding, households suffered a decline in real income and living standards due to the slowdown of economic activity, and higher expenses for health, repairs and fuel. Even though many of the households were able to smooth their consumption, by relying partly on small government transfers and heavy borrowing from private individuals, not everyone was able to maintain adequate calorie consumption during and after the flood. Lower consumption, together with a worsening of the health environment, caused the nutritional status of children in households that were more severely exposed to the flood to deteriorate.
This is not the first time that a serious flood hit the country. Destructive floods are distressingly familiar, though not easily predictable. Bangladesh has suffered at least 10 major floods in the past century. The 1974 flood caused a significant drop in production, and the resulting famine led to 30–100 thousand deaths (del Ninno et al., 2002a, del Ninno et al., 2002b, Ravallion, 1987). But even less severe floods can cause considerable damage: the flash floods that struck again in the summer of 2000 killed more than a thousand people and left millions homeless.
Households may be better prepared to deal with periodically-occurring shocks, such as drought and flood, than with idiosyncratic risks such as unemployment or severe illness (Gertler and Gruber, 1997). On the other hand, repeated shocks can wear one down. Alderman (1994) finds that households in Pakistan had more difficulty in smoothing consumption after successive shocks than after a single shock, and Deaton (1992) finds that the ability of households to maintain consumption diminishes as the autocorrelation in shocks rises. Households can draw on assets to smooth consumption over the course of one disaster; however, a succession of disasters is more difficult to overcome, even if the household learns from experience. Rashid (1999) suggests that some households may not be able to smooth their income and ensure that their children have enough food that would maintain the level of growth. Branca et al. (1993) and Pastore et al. (1993) show that seasonal fluctuations in consumption among Ethiopian households can lead to significant fluctuations in body weight and in work ability. Using repeated cross-sectional observations, Wright et al. (2001) show large seasonal differences in weight among young children in Zimbabwe. Block et al. (2003) find that households were able to maintain the weight of children during the economic crisis in Indonesia. Drawing on data from an earlier flood in Bangladesh, Foster (1995) found that not all households were able to smooth consumption over the crisis: children from credit-constrained poor households suffered relative to others.
In spite of the unprecedented severity of the 1998 flood, not all children suffered; and among those who did suffer, some may have recovered more quickly than others. We begin by comparing the health of children who experienced the flood with those who did not. If we are not able to find a difference in levels between these two groups, flood-exposed households were either able to perfectly smooth consumption over the crisis, or have already been able to recover completely. It is also possible that some flood-exposed households were better able than others to cope and recover. Differences in health within the group of flood-exposed households may be due to the same disparity in the success of household coping mechanisms: either some flood-exposed households smoothed consumption perfectly, or they have been able to recover quickly.
If children in some flood-exposed households have been able to recover, or were able to maintain adequate consumption during and after the flood period, how did they do it? We can postulate three levels of successful coping. First, the child may have his own internal physical resources that encourage health and recovery. He may be well-endowed, genetically, or well-nourished and generally healthy prior to the flood. This child may be better able to cope with physical stress. Second, the household may be better able to cope, independently of outside assistance. It is likely that wealthier households, with greater assets or other resources, are better able to help their children in times of crisis. Finally, the household and its members may be able to draw on community resources, such as informal transfers or credit, or formal assistance from the government or NGOs. We can explicitly test the importance of the latter two coping mechanisms. The first, and arguably most important—the internal resources of the child—we cannot measure directly; but we suggest that they can be affected by some types of external assistance.
We are especially interested in the impact of community resources and public action on health and growth. What actions can be taken to minimize the probability that a household will suffer during a crisis (ex ante), and what can be done to minimize the consequences to household welfare (ex post) once a crisis has occurred? One can imagine two alternative public sector food security programs: one which responds to the crisis, and another which is always present. Given the considerable expense involved in maintaining food security programs in the absence of crises, a great deal of effort has been expended to design distribution systems that can respond quickly to rapidly-deteriorating indicators of food security (Lundberg and Diskin, 1995). Since both types of programs exist in Bangladesh, we can examine their relative efficacy in preventing malnutrition and encouraging recovery.
Both the short and long-term adverse consequences of childhood malnutrition are well known. Malnourished children are more likely to suffer opportunistic infections and more likely to die than well-nourished ones. By one estimate, more than half of all children who die before the age of five are mildly or moderately malnourished (Pelletier et al., 1993). Malnutrition leads to slower cognitive and psychomotor development, greater behavioral problems, and lower educational attainment (Martorell, 1995, Martorell, 1999; Pollitt, 1990, Glewwe et al., 2001, Alderman et al., 2001a, Alderman et al., 2001b). In adults, previous childhood malnutrition is manifest in lower attained body size (Martorell, 1999), reduced labor productivity and earnings (World Bank, 1993), and lower intellectual achievement (Martorell et al., 1992). Childhood malnutrition may also be passed from one generation to the next. Maternal height is a key determinant of low birth weight (Kramer, 1987) and infant mortality (Martorell et al., 1981), as well as maternal mortality (Royston and Armstrong, 1989).
We examine changes in children's health, focusing on growth in standardized height. Although there has been a great deal of research that examines the level of child height and weight, Deolalikar (1996) argues that the stock of height or weight may not be a good indicator of actual health, since it reflects the cumulative influence of past levels of health as well as current inputs. Some studies show that growth rates are predictive of increased mortality risk (Martorell, 1995). Some of the literature examines growth implicitly, for example by inferring growth from an examination of levels as the outcome of a growth process (Pinstrup-Andersen et al., 1995). Among those that examine growth explicitly are Behrman et al. (1992), Foster (1995), Deolalikar (1996) and Hoddinot and Kinsey (2001).
The evidence of “catch-up” growth is mixed, but generally optimistic. Behrman et al. (1992) find that Philippine children are able to recover relatively quickly from seasonal fluctuations in health. Foster (1995) estimates weight growth unconditional on lagged weight, and finds faster growth among children in better-off households or those who had better access to credit markets. Deolalikar (1996) finds almost complete catch-up from low birth weights by the first birthday. However, those results may be biased by sample selection. His sample is based on recall data (as opposed to a prospective study), and sample only children still living. To the extent that low birth-weight babies are more likely to die, the children included in the sample are healthier than the general population of children at birth, and may recover from shocks relatively quickly. Razzaque et al. (1990) find that children born during famine in Bangladesh are more likely to die before age two than children conceived during the famine. They interpret this as evidence that the impact of the famine on child mortality diminishes over time.
In contrast, Hoddinot and Kinsey (2001) present evidence that temporary shocks can have lasting consequences. They examine the impact of drought on height growth among children in Zimbabwe. They find that children aged 12–24 months during the drought lose up to 2 cm of growth in the short term, and that this difference is still evident at least 4 years following the end of the drought. No difference was found among older children. This illustrates the concerns of Martorell (1989), who found that the possibilities for catch-up growth are very limited once a child reaches 3–5 years of age.
Both Foster (1995) and Deolalikar (1996) focus on changes in children's weight, rather than height. Although there is considerable evidence linking child height with subsequent adult health and other outcomes, there is little evidence of the long-term impact of fluctuations in children's weight. This may be because weight fluctuates much more than height in the short-run, and because very short-run changes in weight are not necessarily manifest in long-term height growth.
The use of the standardized measures height-for-age and weight-for-height poses an even greater problem, in that the two measures confound each other: children who are malnourished according to one measure (e.g. height-for-age) are ceteris paribus less likely to be malnourished according to the other (weight-for-height). Shorter children will weigh more, given their height, than taller children. Both numerator and denominator of the latter measure are indicators of nutritional status. It is impossible to determine whether a child is at a particular point on the weight-for-height scale because of changes in his weight or his height. It is true that for a given height, there is some minimum body mass required to maintain healthy physiological functioning. However, to use this as a measure of health requires that we can legitimately assume that the denominator is exogenous to the measure. This assumption is arguably maintainable in the very short run; however, in both the descriptive and analytical sections below, we focus our attention on health mainly in terms of height-for-age.3
Section snippets
The data and descriptive statistics
The dataset used for the study comes from a three-round household survey in seven flood-affected thanas, collected between November 1998 and November 1999.4
Modelling changes in nutritional status
Our comparison of growth rates across children relies on the NCHS standard reference tables (Kuczmarski et al., 2002), which imply that there is an ideal level of height and weight (a “steady-state” level) for a child of a given age and sex. Econometric estimates of the determinants of child anthropometric characteristics, whether in levels or differences, examines why a child's values deviates from the references.
If the levels model is correct, children who have experienced a shock should
The impact of the flood on child growth—empirical results
This section presents the results of the main model. Here we take account of the fact that lagged health status may be measured with error, and other variables, such as the assistance the household receives, may also be endogenous to the health of the child. We now turn to the main results, to see whether growth rates differ across flood exposure. We have seen that flood-exposed children are smaller than those who were not exposed, and we would like to see that they are growing faster than
Conclusions and policy implications
We show that the children exposed to the flood were systematically smaller than those who were not exposed, and that this pattern holds in all three rounds of the survey. In addition, flood-exposed children were not growing more rapidly than their unexposed counterparts after the flood withdrew. Hence, flood-exposed children did not recover from the calamity. This may be due to the lack of proper health care and nutrition, or to the length and severity of the shock itself. Although we do not
Acknowledgements
This research was initiated when Carlo del Ninno was an International Food Research Institute (IFPRI) Research Fellow working on a 5-year Food Management and Research Support Project (FMRSP) financed by USAID/Dhaka. We also gratefully acknowledge the helpful assistance of the FMRSP Dhaka staff. We benefited from comments from Harold Alderman, Emanuel Skoufias and from seminar participants at IFPRI, the 17th International Congress of Nutrition in Vienna, the National Institute of Public Health
References (48)
- et al.
Another look at the instrumental-variable estimation of error-components models
J. Econometr.
(1995) - et al.
Averting a food crisis: private imports and public targeted distribution in Bangladesh after the 1998 flood
Agric. Econ.
(2001) - et al.
Public policy, markets and households coping strategies in Bangladesh: avoiding a food security crises following the 1998 floods
World Dev.
(2003) - et al.
Early childhood malnutrition and academic achievement: a longitudinal analysis
J. Public Econ.
(2001) Saving and economic shocks in rural Pakistan
J. Development Econ.
(1994)- Alderman, H., Hoddinot, J., Kinsey, B., 2001a. Long-term consequences of early childhood malnutrition....
- et al.
Attrition in longitudinal household survey data. Some tests from three developing countries samples
Demogr. Res.
(2001) - et al.
Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations
Rev. Econ. Stud.
(1991) - Behrman, J., Deolalikar, A., Lavy, V., 1992. Dynamic decision rules for child growth in rural India and the...
- et al.
Macro shocks and micro outcomes: child nutrition during Indonesia's crisis
Econ. Hum. Biol.
(2003)
The nutritional impact of seasonality in children and adults of rural Ethiopia
Eur. J. Clin. Nutr.
Saving and income smoothing in the Côte d’Ivoire
J. Afr. Econ.
Child nutritional status and child growth in Kenya: socioeconomic determinants
J. Int. Dev.
The 1998 floods and beyond: towards comprehensive food security in Bangladesh
Prices, credit markets and child growth in low income rural areas
Econ. J.
Cited by (78)
Role of in-situ and ex-situ livelihood strategies for flood risk reduction: Evidence from the Mayurakshi River Basin, India
2022, International Journal of Disaster Risk ReductionCitation Excerpt :At the post-flood period, the income level declines with increasing expenses for health issues, repairing the damages to buildings and infrastructure. Even victims failed to manage proper nutrition which induces different health issues during and after the flood [49]. The expenses regarding food (per capita per month) range from INR 242 (Burwan) to INR 617 (Kandi) in the non-flood periods.
Impact pathways from climate services to SDG2 (“zero hunger”): A synthesis of evidence
2022, Climate Risk ManagementCitation Excerpt :A substantial body of literature, including reviews and meta-analyses (Asmall et al., 2021; Belesova et al., 2019; Brown et al., 2020; Cooper et al., 2019b; Delbiso et al., 2017), documents near-term impacts of weather and seasonal climate extremes on food consumption; the prevalence of food-insecure or undernourished individuals; and health impacts expressed as wasting (i.e., low weight-for-height), stunting (i.e., low height-for-age), underweight (i.e., low weight-for-age) and mid-upper arm circumference (MUAC). While the most frequently reported climatic trigger is drought (Amare et al., 2018; Bahru et al., 2019; Bauer and Mburu, 2017; Belesova et al., 2019; Cooper et al., 2019b; Delbiso et al., 2017; Hoddinott and Kinsey, 2001), studies also document impacts of high and low temperature extremes (Asfaw and Maggio, 2018; Brown et al., 2020; Hagos et al., 2014; Randell et al., 2020), flooding (Akukwe et al., 2020; del Ninno and Lundberg, 2005; Muttarak and Dimitrova, 2019), and excess rainfall (Cooper et al., 2019a). A meta-analysis of 90 studies of factors associated with child malnutrition found evidence of statistically significant links between drought and underweight, between excess rainfall and wasting, and between extreme temperature and stunting (Brown et al., 2020).
Climate change, social vulnerability and child nutrition in South Asia
2021, Global Environmental ChangeCitation Excerpt :We also know that children’s vulnerability to climate shocks does not remain constant throughout their early years (Shrimpton et al., 2001, Victora et al., 2010, Headey et al., 2015). In some cases, partial or complete catch-up growth among affected children has been shown to reduce the harmful effects of shocks over time (Hoddinott and Kinsey, 2001; del Ninno and Lundberg, 2005), though this phenomenon is sometimes limited to children belonging to households with more access to wealth and community support. At the same time, evidence shows that exposure to extreme weather during the first 1,000 days after conception can have adverse developmental effects that last into adulthood and are linked to educational, financial, and physical deficits (Rayco-Solon et al., 2005; Alderman et al., 2006; Ramakrishnan et al., 2012; Black et al., 2013; Nobles and Hamoudi, 2019).
Flooding and child health: Evidence from Pakistan
2021, World DevelopmentCitation Excerpt :They use river height to proxy for flooding intensity. Del Ninno and Lundberg (2005) use three rounds of data collected after the same flood to show that the children exposed to flooding were subsequently shorter, with no evidence of catch-up growth over time. They use self- reported data on flood exposure.
Climate exposures and child undernutrition: Evidence from Indonesia
2020, Social Science and MedicineCitation Excerpt :Previous studies have directly examined the consequences of environmental change for child nutritional status in Mongolia (Groppo and Kraehnert, 2016), Nepal (Shively et al., 2015), Bangladesh (Del Ninno and Lundberg, 2005), India (Kumar et al., 2016), Mexico (Skoufias and Vinha, 2012), Ecuador (Rosales-Rueda, 2018), and several countries in Sub-Saharan Africa (Hoddinott and Kinsey, 2001; Davenport et al., 2017). A majority of studies have focused on the consequences of extreme events such as drought and flooding (Hoddinott and Kinsey, 2001; Del Ninno and Lundberg, 2005; Rabassa et al., 2014; Groppo and Kraehnert, 2016; Kumar et al., 2016; Rosales-Rueda, 2018). The results have consistently shown large negative effects on child nutritional status, supporting the relevance of the pathways described above.
- 1
Tel.: +1 202 473 2442.