Ibrahim Worku, Mekdim Dereje, and Bart Minten Ethiopia’s food economy is transforming fast. To better understand this ongoing process, we analyze changes in food consumption that have happened over the period between 1996 and 2011, relying on four rounds of nationally representa- tive household data. The main findings are as follows: First, the share of food in overall expenditures is declining. Second, quantities consumed per capita are increasing. Third, the relative importance of cereals is on the decline. Fourth, there are large differences in the cereal basket of households between rural and urban areas, indicating the importance of in- creasing urbanization on the food economy. Fifth, cereals – and in particular maize – remain the largest source of calories in the food basket. Sixth, there is a relative shift in consumption to more expensive foods. Seventh, purchased foods are seemingly becoming more important. Finally, there are strong differences in food consumption by income level, but all levels exhibit changes over time. This diet transformation has important implications for the food security debate and for agricultural and food policy in the country. ESSP RESEARCH NOTE 46 • November 2015 Synopsis: Diet transformation in Ethiopia Introduction Over the last decade, rapid changes have occurred in Ethiopia’s food economy. We explore these changes over this period which is characterized by high economic growth rates, making the coun- try one of the fastest growing economies in the world. To better understand this rapid transformation, the link between food de- mand, urbanization, and income levels in particular is examined, as this might expose a trend in food consumption patterns, espe- cially within a country where urbanization is encouraged and at- taining middle-income status by the year 2025 is described as an objective in several policy documents. Table 1. Food consumption and real per capita expenditures, by food category, Birr/person/year 1996 2000 2005 2011 Birr Share (%) Birr Share (%) Birr Share (%) Birr Share (%) Teff 72 11.5 76 12.9 57 8.9 56 7.5 Wheat 45 7.2 51 8.6 57 8.9 55 7.4 Barley 29 4.7 22 3.8 28 4.4 18 2.4 Maize 63 10.0 64 10.9 55 8.6 58 7.7 Sorghum 45 7.2 37 6.2 52 8.1 37 5.0 Other cereals 18 3.0 19 3.2 11 1.7 11 1.4 Processed cereals 13 2.0 11 1.9 21 3.2 34 4.6 All cereals 285 45.7 281 47.6 281 43.8 269 36.0 Pulses 48 7.6 57 9.6 50 7.7 71 9.5 Oilseeds 2 0.3 2 0.3 1 0.2 1 0.2 Animal products 47 7.5 45 7.7 56 8.7 81 10.8 Oil & fat 29 4.6 22 3.6 25 3.9 50 6.6 Vegetables & fruits 23 3.7 26 4.3 30 4.6 48 6.4 Pepper 30 4.9 24 4.1 17 2.6 50 6.6 Enset/kocho 32 5.1 45 7.7 29 4.4 31 4.2 Coffee/tea/chat 62 9.9 41 6.9 49 7.7 68 9.1 Root crops 17 2.7 20 3.4 20 3.1 13 1.8 Sugar & salt 15 2.5 12 2.0 12 1.9 20 2.7 Other foods 36 5.7 16 2.7 72 11.3 46 6.2 Total food 625 100.0 590 100.0 643 100.0 748 100.0 Food versus non-food Food 625 51.6 590 62.8 643 54.1 748 47.9 Non-food 585 48.4 349 37.2 546 45.9 812 52.1 Total 1210 100.0 939 100.0 1189 100.0 1560 100.0 To analyze consumption patterns, we rely on the Ethiopian Household Consumption and Expenditure Survey (HICES) dataset from four past rounds – 1995/96, 1999/00, 2004/05 and 2010/11. These data were collected by the Central Statistical Agency (CSA). In total, 11,678, 17,320, 21,560 and 27,831 households were in- terviewed over the four survey rounds, respectively. Findings 1. The share of food in overall expenditures is declining The share of non-food items in the total consumption basket of Ethiopian households has increased significantly over time. In 2000, the share of non-food consumption expenditures accounted for 37.2 percent of the total. Over the following decade, this type of expenditure grew rapidly, and its share in total household ex- penditures surged to 52.1 percent in 2011 (Table 1). Such in- creases of non-food expenditures in total consumption are typical of transforming and improving economies, and imply significant improvements in welfare in the country. This is because greater income gives rise to a shift in people’s expenditure from food to non-food products. 2. Quantities of food consumed per capita are increasing Per capita total quantity of food consumed has increased signifi- cantly during the last decade (Table 2). Consumption increased from 247 kg per capita in 1996, to 293 kg in 2000, to 361 kg in 2011, or an increase of 23 percent over the last decade. The quan- tities of cereals consumed have shown much less growth. Here, consumption of cereals grew from 141 kg per capita in 2000 to 155 kg per capita in 2011, an increase of 10 percent. Consistent with this trend, expenditures on food have grown in real terms in the last two surveys compared to 2000. Per capita food expenditures in 2011 were 19 percent higher than in 2000. 3. The relative importance of cereals is on the decline Overall, the share of cereals in total household food expenditures is declining. While the share made up 47.5 percent of expenditures in 2000, this had declined to 35.8 percent ten years later. Growth in the non-cereal food share was recorded in a number of catego- ries. Animal products, for example, showed increasing importance This research has benefited from funding provided by the Feed the Fu- ture Innovation Laboratory for Food Security Policy, funded by the Bureau of Food Security of the United States Agency for International Development Table 2. Food consumption, by food category, kg/capita/year 1996 2000 2005 2011 kg Share (%) kg Share (%) kg Share (%) kg Share (%) Teff 26 10.5 30 10.3 26 8.0 26 7.3 Wheat 21 8.6 24 8.3 30 9.1 25 6.9 Barley 15 5.9 10 3.3 13 3.9 10 2.7 Maize 34 14.0 37 12.6 38 11.6 51 14.2 Sorghum 18 7.1 22 7.7 32 9.9 28 7.8 Other cereals 9 3.5 14 4.7 5 1.6 7 2.0 Processed cereals 4 1.8 4 1.5 6 2.0 7 2.0 All cereals 127 51.4 141 48.3 150 46.2 155 43.0 Pulses 20 8.1 16 5.4 17 5.1 18 5.0 Oilseeds 1 0.3 0 0.2 0 0.1 0 0.1 Animal products 14 5.8 12 4.1 15 4.6 17 4.6 Oil & fat 2 0.8 1 0.5 2 0.7 4 1.2 Vegetables & fruits 27 11.0 26 9.0 34 10.5 36 10.0 Pepper 3 1.4 2 0.6 2 0.6 4 1.2 Enset/kocho 12 4.7 56 19.2 41 12.8 47 13.0 Coffee/tea/chat 9 3.7 7 2.5 8 2.5 12 3.4 Root crops 13 5.3 22 7.5 27 8.3 24 6.7 Sugar & salt 7 2.7 5 1.7 5 1.7 8 2.1 Other foods 12 4.7 3 0.9 23 7.0 35 9.7 Total food 247 100.0 293 100.0 324 100.0 361 100.0 in consumption rates over time. While the share of these prod- ucts is still relatively low, this has grown from 7.6 percent of food expenditures in 2000 to 10.8 percent in 2011. These patterns re- flect Bennett’s law that describes a relative decline in starchy sta- ples and an increase in animal proteins with an increase in in- come. The share of fruits and vegetables also increased over that period, from 4.3 percent to 6.4 percent. Significant positive growth is also seen in the ‘oil and fat’, ‘pepper’ and ‘cof- fee/tea/chat’ categories. In contrast, we see a decline in the root crops and enset/kocho categories. Cereal expenditures made up 36.0 percent of the value of the total food consumption basket in 2011 (Table 1) but comprised 43.0 percent of the quantity consumed (Table 2). This indicates that cereals are relatively lower in cost. In contrast, animal prod- ucts constituted 10.8 percent of expenditures and 4.6 percent of the quantities consumed in 2011. These animal products are the most expensive food items in the consumption basket. The cate- gories ‘roots and tubers’ and ‘enset/kocho’ illustrate the opposite pattern, as these are a relative cheap food category. The most important cereal expenditures are those for teff, wheat, and maize. These crops accounted for 7.5, 7.4, and 7.6 percent, respectively, of food expenditures in 2011. Over time, some minor shifts within the consumption of cereals are ob- served. For example, the share of expenditures of sorghum in ce- real expenditures was 6.3 percent in 2000 and 8.1 percent in 2005, but this declined to 4.9 percent in 2011. Compared to 2000, the share of maize in cereal expenditures has decreased as well. However, in quantity terms, maize is still by far the most im- portant cereal consumed. Within the cereal category, we note the increase of processed cereals, from 1.9 to 4.6 percent of total food expenditures, still relatively low compared to other African countries. 4. Large differences in the cereal basket in urban versus rural areas The importance of cities is rapidly increasing. While the share of the urban population in Ethiopia is relatively small, the im- portance of cities, however, has rapidly increased and is expected to further increase in the future. This has likely implications on the food economy as urban residents usually do not grow their own food and tend to have different food baskets. Average per capita expenditures are significantly higher in ur- ban than in rural areas, and the share of non-food expenditures is also significantly higher in urban (62.3 percent) than in rural (48.8 percent) areas. Interestingly, there are almost no differ- ences in the share of cereals in the food consumption basket, and the quantities of cereals consumed are also at similar levels (Ta- ble 3). However, within the cereal category, consumption pat- terns differ significantly. Rural consumers consume significantly more sorghum (32 kg versus 12 kg) and maize (58 kg versus 18 kg). In contrast, the share of teff in the urban food consumption basket is significantly higher than in rural areas, more than twice as high. Moreover, urban consumers eat 59 kg of teff per year, almost three times the level consumed in rural areas. Consumption of animal products is significantly higher in ur- ban areas. We also note significant differences in the consump- tion of enset/kocho which is almost exclusively eaten in rural ar- eas (55 kg per capita on average, compared to 6 kg in urban ar- eas). The root crops are also a more important source of food in rural areas than in urban areas (26 kg and 16 kg per capita in rural and urban areas, respectively). Table 3. Food consumption in 2011, urban versus rural, by food category, kg/capita/year Urban Rural Total kg Share (%) kg Share (%) kg Share (%) Teff 59 18.4 20 5.4 26 7.3 Wheat 35 10.9 23 6.2 25 6.9 Barley 4 1.1 11 3.0 10 2.7 Maize 19 5.8 58 15.7 51 14.2 Sorghum 12 3.6 32 8.6 28 7.8 Other cereals 3 1.0 8 2.2 7 2.0 Processed cereals 23 7.0 4 1.1 7 2.0 All cereals 153 47.9 156 42.2 155 43.0 Pulses 18 5.6 18 4.9 18 5.0 Oilseeds 0 0.1 0 0.1 0 0.1 Animal products 19 6.1 16 4.4 17 4.6 Oil & fat 9 2.9 3 0.8 4 1.2 Vegetables & fruits 47 14.8 34 9.2 36 10.0 Pepper 5 1.6 4 1.1 4 1.2 Enset/kocho 6 2.0 55 14.9 47 13.0 Coffee/tea/chat 8 2.5 13 3.6 12 3.4 Root crops 16 5.1 26 7.0 24 6.7 Sugar & salt 11 3.5 7 1.9 8 2.1 Other foods 25 7.9 37 10.0 35 9.7 Total food 320 100.0 369 100.0 361 100.0 5. Cereals, in particular maize, stay the biggest contributor to calorie consumption Quantities consumed were further converted to calories per adult equivalent using the standard conversion rates from food compo- sition tables for Ethiopia (Table 4). A consistent increase in per capita calorie consumption is seen over the years, reflecting the improving food security situation in the country. Average calorie consumption was only 2,321 kcal per day per adult equivalent in 1996, but this measure reached 3,001 kcal in 2011. Cereals are the major contributors to total calorie consump- tion, contributing 61.8 percent of all the calories in the food bas- ket in 2011. While their absolute level of consumption has in- creased over time, their share in total consumption, however, has decreased from 66.5 percent in 1996. With regard to contribution of calories, maize is the most important source of calories. It ac- counts for 20 percent of the average calories consumed per adult equivalent. Sorghum accounts for 12 percent and teff and wheat are of equal importance (between 10 and 11 percent). Barley and other cereals are of less importance. While processed cereals ac- count for almost 5 percent of expenditures, they contribute rela- tively less towards calories, with 2 percent of calories provided by this category of cereals. Table 4. Calorie consumption per adult equivalent per day 1996 2000 2005 2011 kcal. Share (%) kcal. Share (%) kcal. Share (%) kcal. Share (%) Teff 320 13.8 372 13.6 316 11.1 323 10.8 Wheat 275 11.8 316 11.5 364 12.8 301 10.0 Barley 137 5.9 92 3.4 113 4.0 121 4.0 Maize 455 19.6 490 17.9 441 15.5 608 20.3 Sorghum 227 9.8 289 10.6 363 12.7 353 11.8 Other cereals 77 3.3 169 6.1 63 2.2 91 3.0 Processed cereals 53 2.3 57 2.1 86 3.0 57 1.9 All cereals 1,544 66.5 1,785 65.1 1,745 61.3 1,854 61.8 Pulses 255 11.0 205 7.5 203 7.1 214 7.1 Oilseeds 11 0.5 7 0.3 7 0.2 5 0.2 Animal products 75 3.2 64 2.3 76 2.7 59 2.0 Oil & fat 57 2.4 39 1.4 63 2.2 130 4.3 Vegetables & fruits 64 2.8 65 2.4 76 2.7 62 2.1 Pepper 10 0.4 5 0.2 5 0.2 52 1.7 Enset/kocho 67 2.9 323 11.8 235 8.2 306 10.2 Coffee/tea/chat 47 2.0 39 1.4 41 1.5 60 2.0 Root crops 71 3.1 122 4.4 149 5.2 104 3.5 Sugar & salt 90 3.9 67 2.4 72 2.5 58 1.9 Other foods 31 1.3 21 0.8 173 6.1 97 3.2 Total food 2,321 100.0 2,742 100.0 2,846 100.0 3,001 100.0 6. A relative shift in consumption to more expensive foods Comparing expenditures with calorie consumption allows for the price per calorie of food consumed to be calculated. Figure 1 shows that the average price per calorie differs significantly be- tween food groups. When the prices for 2011 are considered, the prices of cereals are significantly lower than those of almost all other categories. These other categories are therefore usually called “high-value”. Animal products carry the highest price for calories with a calorie price that is eight times as high as the av- erage price paid for cereals. The price for fruit and vegetables is also relatively high as it has a price tag that is about five times as high per calorie than for cereals. The price for other foods – ex- cept for enset/kocho and root crops – is also significantly higher than the average. Within the cereal category, lowest calorie prices are found for maize and sorghum, which are significantly below those for teff and wheat. Calories from processed cereals have a significantly higher average price. Figure 1. Real price paid per kilocalorie, 2011 The shift that is seen over time in food preferences shows a reduction in consumption of foods comprising low-priced calo- ries, while there is an increase in consumption of more expensive ones. The average price that consumers pay per calorie has in- creased by 8 percent over the last decade. This is mostly driven by a shift to more expensive commodities. 7. Purchased foods are becoming more important The HICES survey required respondents to indicate how they paid for specific foods. The results of these answers are illustrated in Figure 2. We see large differences in source of payment between urban and rural areas. Consumption of own agricultural products accounts for 42 percent of total food expenditures in rural areas. This reflects the high level of subsistence of the rural Ethiopian economy. However, this number is significantly lower than what is usually assumed. 34 percent of food consumed by rural house- holds is paid for through the sales of agricultural products. 76 per- cent of food expenditures in rural areas are therefore directly paid for by agricultural income, illustrating less importance at- tached to off-farm income. Remittances, sales of non-agricultural products, and wage income account for 9 percent, 7 percent, and 4 percent, respectively, of total food expenditures in rural areas. Figure 2. Sources of payments for food, rural and urban, 2011 In urban areas, most food expenditures are paid for by income from wages (40 percent). Second comes the sale of non-agricul- tural products (28 percent). Remittances make up 15 percent of all food expenditures and apparently are a more important source of income in urban centers than in rural areas. Food ex- penditures made through “government donation and social secu- rity” are relatively less important. They make up 2.3 percent and 1.6 percent of total food expenditures in urban and rural areas, respectively. In rural areas, the share provided by the govern- ment seems mostly related to the Productive Safety Net Program (PSNP) as well as by emergency aid, as indicated by the relatively large share of government donations in the case of wheat con- sumption (5.5 percent) that is paid for that way 8. Strong differences in food consumption by income level, but all income levels exhibit changes over time A number of food consumption patterns can be distinguished with increasing income and economic development:  Processed and ready-to-eat foods take off;  Cereals become less important;  The share in food consumption baskets of high-value crops, such as fruit and vegetables, dairy and animal products, and fish, increases. Comparing the differences in consumption patterns of richer and poorer households helps to identify how transformation of food systems will shape food economies. In an effort to under- stand these patterns in Ethiopia, all households in the HICES sur- vey of 2011 were ranked by wealth quintile, from the poorest quintile 1 to the richest quintile 5. The shares of different con- sumption categories were then calculated. Strong differences in the composition of consumption baskets are seen over wealth quintiles. While food expenditures make up 55.0 percent of total consumption expenditures for the poorest quintile, this declines to 38.6 percent for the richest one. The five Table 5. Share of expenditures by food category, by wealth quintile, 2011 Q1 (poorest) Q2 Q3 Q4 Q5 (richest) Total Teff 4.4 6.3 7.7 8.5 9.1 7.5 Wheat 6.5 6.9 7.8 7.8 7.6 7.4 Barley 3.4 3.2 2.8 2.1 1.1 2.4 Maize 13.0 10.4 8.8 6.5 2.6 7.7 Sorghum 6.9 6.4 6.3 4.8 1.8 5.0 Other cereals 1.9 2.0 1.6 1.2 0.8 1.4 Processed cereals 2.6 2.6 3.1 4.3 8.9 4.6 All cereals 38.6 37.8 38.0 35.1 32.0 36.0 Pulses 10.1 10.0 10.7 9.7 7.6 9.5 Oilseeds 0.2 0.2 0.2 0.2 0.1 0.2 Animal products 6.6 7.6 8.1 11.4 17.6 10.8 Oil & fat 4.8 6.0 6.1 7.1 8.3 6.6 Vegetables & fruits 6.4 6.4 6.2 6.3 6.7 6.4 Pepper 7.2 7.3 6.9 6.6 5.6 6.6 Enset/kocho 6.4 5.8 4.0 4.2 1.7 4.2 Coffee/tea/chat 9.1 9.5 9.6 9.1 8.2 9.1 Root crops 3.6 2.1 1.6 1.3 1.0 1.8 Sugar & salt 2.4 2.4 2.5 2.9 3.0 2.7 Other foods 4.7 4.9 6.1 6.0 8.3 6.2 Total food 100.0 100.0 100.0 100.0 100.0 100.0 Food versus non-food Food 55.0 54.6 51.9 48.2 38.6 47.9 Non-food 45.0 45.4 48.1 51.8 61.4 52.1 Total 100.0 100.0 100.0 100.0 100.0 100.0 major cereals make up 34.2 percent of the poorest quintile’s food expenditures. This declines to 22.2 percent for the richest quintile (Table 5). Notably, the share is relatively stable for the poorest three quintiles and drops off only for quintiles 4 and 5, suggesting that transformation in the content of food baskets away from ce- reals only starts to occur as household wealth rises to the levels of the richest two quintiles. Animal foods for the richest house- holds comprises 17.6 percent of the value of their food basket, yet this is only 6.6 percent for the poorest. Overall, the richer con- sumers become, the more they spent on food and the more food they consume. Cereal consumption patterns differ considerably by wealth quintile. Maize and sorghum are the cereals that are typically con- sumed more by the poor than by the rich. 13.0 percent of all food expenditures of households in the poorest quintile is spent on maize. This compares to 2.6 percent of the expenditure of the richest. For sorghum, these numbers are 6.9 and 1.8 percent, re- spectively. The consumption of barley is low overall, but its share also decreases as wealth increases. There is relatively little varia- tion for wheat consumption across wealth quintiles. To further explore patterns of food consumption expendi- tures between wealth quintiles and changes in consumption be- havior patterns, the left graph in Figure 3 illustrates how the share of food in the total consumption basket has evolved over time by quintile. The figure illustrates that the reduction in the food share has been consistent over quintiles. While food ex- penditures made up 71 percent of total expenditures of the poor- est quintile in 2000, this decreased to 54 percent in 2011. Similar reductions over time are seen for all quintiles. In the right graph in Figure 3, the share of cereals in food expenditures is shown for the five quintiles over the last three surveys. The share of cereals for the poorest quintile was 47 percent in 2000 but declined to 38 percent in 2011. Again, there has been a consistent decrease of the share of cereals for all quintiles in the consumption basket, even though the average quantities consumed increased over time. Overall, these findings seemingly indicate that economic growth in the last decade in Ethiopia has been equitable and seems to have benefited a majority of the population. Figure 3. Share of food in total consumption expenditures (left) and share of cereals in food expenditures (right), by wealth quintile, 2000, 2005, and 2011 Implications These findings lead to a number of policy implications. First, agri- cultural policy in Ethiopia has been successful in improving productivity and availability of cereals in the country, seemingly contributing to improved food security for large parts of the pop- ulation. However, as shown here, there is still a lack of diversity in diets which seems to be linked to high levels of stunting in the country. More emphasis on diversification in agricultural produc- tion as well as on diversification of diets seems required. Second, agricultural markets play an increasing role in provid- ing consumers with the food that they require. By gaining a better understanding of how these markets might assure adequate, di- verse diets, they could help satisfy current needs. This is espe- cially sought given the large distrust that usually exist with re- spect to these markets. Further understanding of producer, wholesale and retail markets; agricultural processing; trade logis- tics; and the role of each of these factors in shaping food prices and consumption patterns is required in order to design appro- priate policies and interventions to improve diets in Ethiopia. Third, while there has been significant improvement in the amount and types of calories consumed by average consumers, there is still a significant section of the population that does not have access to an adequate diet. Hence, a further strengthening of targeted safety nets seems essential to improve the food secu- rity of deprived people in order to improve their nutrition, health, and well-being. 0 10 20 30 40 50 60 70 80 q1 q2 q3 q4 q5 Sh ar e o f fo o d in t o ta l e xp e n d it u re s (% ) 2011 2005 2000 0 10 20 30 40 50 q1 q2 q3 q4 q5 Sh ar e o f ce re al s in f o o d e xp e n d it u re s (% ) 2011 2005 2000 mailto:ifpri@cgiar.org http://www.ifpri.org/ mailto:info@edri-eth.org http://www.edri-eth.org/ mailto:mahlet.mekuria@cgiar.org http://essp.ifpri.info/