Using Household Consumption and Expenditure Surveys to Understand Nutrition Transition in Ethiopia

Ethiopia is a landlocked country located in the northeastern part of Africa. It is the second most populous country in the continent after Nigeria, and one of the fastest growing economies in the world (World Bank, 2018a). Although it is also one of the world’s poorest countries, economic growth in the past decade has contributed to impressive poverty reduction in both urban and rural areas. The share of the population below the poverty line fell from 61.2% in 1999 to 27.3% in 2015 (World Bank, 2018b). Per Bennett’s Law (Bennett, 1941), improvements in living standards are usually accompanied by a shift in diet composition from starchy staple foods to higher value products (meat, dairy products, fruits and vegetables) (Gouel et al., 2018; Masters et al., 2013; Popkin et al., 2003). The following case study, based on work by Worku et al. (2017), illustrates an example of using national Household Consumption and Expenditure Survey (HCES) data to characterize changes, and associated drivers, in Ethiopia’s dietary patterns over a 15-year period of rapid economic growth. The following paragraphs summarize the methods and findings from Worku et al. (2017) and highlight the strengths and weaknesses of using HCES data to assess household food consumption for their research purpose.

Worku et al. (2017) analyzed four Ethiopian Household Consumption and Expenditure Surveys from 1995/96, 1999/00, 2004/05, and 2010/11 to identify changes in household expenditure levels and food consumption patterns. The Ethiopia HCES offered a detailed consumption and expenditure module; the food list collected data on 275 foods over a 7-day recall period. The authors calculated the household food expenditure share of total spending to assess how food and non-food expenditures changed with increases in total expenditures. They then classified food expenditures by food group (expressed in constant 1996 birr per adult equivalent per year) in order to assess how expenditures by type of food had evolved over the time series.

To examine changes in food consumption quantities over time, the authors transformed data on amounts consumed from local measurement units into standardized gram equivalents and then further converted the amounts consumed to calories using the energy composition data reported in the Ethiopian Food Composition Table (FCT) (EHNRI, 1968-1997). After this step, the authors calculated average dietary energy consumption expressed as kcal per adult equivalent per day. To complement these data and look at the cost of calories across different food items over time, the authors derived the mean real cost of a calorie per day (birr/kcal) by dividing per adult equivalent calorie intakes with per adult equivalent expenditures. To examine the role of income level in expenditure and consumption, households in each survey sample were ranked by expenditure quintile and the expenditure share of each quintile was computed. A formal “decomposition analysis” modeled the proportion of change in calorie consumption explained by income, prices and other factors.

Worku et al.’s (2017) findings revealed that total expenditure levels in Ethiopia increased by 56% from 1996 to 2011, and household food expenditure share declined by an average of 1.1% per year over the 15-year period.  Correspondingly, calorie consumption increased by an impressive 804 kcals to 3,001 kcals per adult equivalent by 2011. The composition of the average household food basket shifted as well, with cereals and other starchy staples comprising an increasingly smaller share of the total expenditures over the 15 years. Notably, the absolute quantity (KG) of cereals consumed increased by 29% and roots/tuber consumption nearly doubled, while the calorie contribution of starchy staples to the diet, expressed as a percent of total calories, was nearly constant over the time series, at 72.6% in 1996 and 75.5% in 2010. High-value foods such as animal products, fruits, vegetables and processed foods formed a greater share of food expenditures in the later surveys but remained at low absolute levels in terms of quantity and calorie share relative to other countries in sub-Saharan Africa and international guidance (Ruel et al., 2005 as cited in Worku et al., 2017). These trends suggest that, though Ethiopia has begun a nutrition transition, diets remain relatively less diverse than would be optimal for health.

The Worku et al. (2017) results also showed differences in urban and rural diets, with average per adult equivalent food expenditures and share of nonfood expenditures higher in urban areas, but actual amounts consumed (KG) higher in rural locations, possibly due to differences in food costs. In addition, urban households were found to allocate a much larger share of their food budget to animal source foods, oils and fats, and fruits and vegetables than their rural counterparts. The authors found that the positive changes they documented in calorie consumption were mostly (84%) explained by improvements in total household expenditures (a proxy for income), and the remainder was primarily explained by decreases in calorie costs.      

The authors caution that these results are from observational data and, therefore, one must be careful in drawing strong conclusions about causality. Yet, at minimum, the descriptive findings are of considerable policy relevance for Ethiopia. The authors advise implementing food policies that emphasize further diversification of agricultural production and diets, along with improving the potential for agricultural markets to reach consumers with adequate, affordable, and nutritionally diverse food choices.

This case study is an example of how existing HCES data can be used to assess consumption trends over time. Household average dietary energy consumption was measured per adult equivalent, which provides a more accurate picture than a per capita measure since it takes into account household age and sex composition. HCES have been shown to be a valuable data source for analyzing household level food security and diets (Russel et al., 2018; Fiedler et al., 2016; Smith et al., 2014; Fiedler et al., 2012) for many reasons: (1) HCES samples are typically nationally, and sometimes also regionally, representative; (2) the data are collected fairly frequently, i.e., every 3-5 years; and (3) HCES include a wide range of data on diet determinants and outcomes (e.g. socioeconomic status, health, education, production), allowing for various analytical options.

However, HCES consumption and expenditure modules also have some weaknesses, making many HCES surveys not as useful as individual-level dietary surveys for certain types of analyses. For example, most HCES do not collect information on food consumed away from home. In countries facing a nutrition transition, an increasingly large share of diets tend to derive from food purchased and consumed outside the household, and this information is often not captured in HCES. Furthermore, many HCES only measure ‘apparent consumption’, which is based on data about food purchases, not actual consumption. Food lists are not always very detailed and may not be disaggregated to the level of foods of interest for certain types of analyses (e.g. identifying vehicles for fortification, gauging coverage of fortified foods, or matching foods to nutrient composition information in a FCT to calculate household nutrient intakes) (Bell et al., 2019). Furthermore, recall periods vary greatly by survey, from 1 to 365 days (Smith et al., 2014), raising concerns about recall bias in those with lengthier recall periods (Beegle et al., 2010). The methods used to quantify consumption are not typically implemented as precisely or accurately as those used in 24-hour Dietary Recalls (24HRs), and the household-level data results cannot be used to draw conclusions about population subgroups, such as women or children. Although collected infrequently and often in smaller samples, data from 24HR and Food Frequency Questionnaires offer more accurate estimates of individual food and nutrient intake, and should be used when possible to assess dietary trends. Recently, a growing number of low- and middle-income countries have prioritized investments in nationally representative individual-level dietary data. An approach that could enable the integration of individual dietary data and HCES results would be to collect the nationally representative 24HR data from the same sampled household (or a sub-sample) interviewed for the routine HCES.

In the Ethiopia study, the researchers took advantage of available national level HCES data to study trends over time in household consumption and their relationship with economic growth in Ethiopia. This analysis provides policy makers with a better understanding of food system transformations taking place in the country, in order to support the design of appropriate policies to improve the country’s food and nutrition security.

Citations:

1. Beegle, K., Joachim D. W., Friedman, J., Gibson, J. (2010). Methods of household consumption measurement through surveys: Experimental results from Tanzania. The World Bank. https://doi.org/10.1596/1813-9450-5501. Retrieved from https://elibrary.worldbank.org/doi/pdf/10.1596/1813-9450-5501
2. Bell, W., Coates, J. C., Rogers, B. L., & Bermudez, O. I. (2019). Getting the food list ‘right’: An approach for the development of nutrition-relevant food lists for household consumption and expenditure surveys. Public Health Nutrition, 22(2), 246–256. https://doi.org/10.1017/S1368980018002847
3. Bennett, M.K. (1941). International Contrasts in Food Consumption. Geographical Review. 31 (3): 365–376. doi:10.2307/210172. ISSN 0016-7428. JSTOR 210172.
4. EHNRI, 1968–1997. Food Composition Table for Use in Ethiopia Part III. Ethiopian Health and Nutrition Research Institute (EHNRI), Addis Ababa.
5. Fiedler, J., Mwangi, D.M. (2016). Using Household Consumption and Expenditure Surveys to Make Inferences about Food Consumption, Nutrient Intakes and Nutrition Status: How Important Is It to Adjust for Meal Partakers? IFPRI Discussion Paper No. 1571.
6. Fiedler, J., Lividini, K., Bemudez, O., & Smith, M. (2012). Household Consumption and Expenditure Surveys (HCES): A primer for food and nutrition analysts in low- and middle-income countries. Food and Nutrition Bulletin, 33(3) suppl, S170-S184.
7. Gouel, C., Guimbard, H. (2018). Nutrition transition and the structure of global food demand. American Journal of Agricultural Economics. 101 (2): 383–403. doi:10.1093/ajae/aay030
8. Masters, W.A., Andersson Djurfeldt, A., De Haan, C., Hazell, P., Jayne, T., Jirström, M., Reardon, T. 2013. Urbanization and farm size in Asia and Africa: Implications for food security and agricultural research. Global Food Security, 2(3), 156-165. https://doi.org/10.1016/j.gfs.2013.07.002
9. Popkin, B.M. (2003). The nutrition transition in the developing world. Dev Policy Rev, 21(5-6), 581-597.
10. https://doi.org/10.1111/j.1467-8659.2003.00225.x
11. Ruel, M.T., Minot, N., Smith, L., 2005. Patterns and Determinants of Fruit and Vegetable Consumption in Sub‐Saharan Africa: A Multicountry Comparison. World Health Organization, Geneva.
12. Russel, J., Lechner, A., Hanich, W., Delisle, A. (2018). Assessing food security using household consumption expenditure surveys (HCES): a scoping literature review. Public Health Nutrition, 21(12), 2200-2210. https://doi.org/10.1017/S136898001800068X
13. Smith, L., Dupriez, O., Troubat, N. (2014). Assessment of the Reliability and Relevance of the Food Data Collected in National Household Consumption and Expenditure Surveys. http://www.ihsn.org/node/683
14. Worku, I.H., Dereje, M., Minten, B., & Hirvonen, K. (2017). Diet transformation in Africa: the case of Ethiopia. Agricultural Economics, 48, 73-86. https://doi.org/10.1111/agec.12387
15. World Bank. (2018a). The World Bank in Ethiopia. Retrieved from https://www.worldbank.org/en/country/ethiopia/overview#1
16. World Bank (2018b). Data: Indicators. Retrieved from https://data.worldbank.org/indicator/SI.POV.DDAY?locations=ET