MEASURING THE AFFORDABILITY OF
NUTRITIOUS DIETS IN AFRICA: PRICE INDEXES
FOR DIET DIVERSITY AND THE COST OF
NUTRIENT ADEQUACY
WILLIAM A. MASTERS, YAN BAI, ANNA HERFORTH, DANIEL B. SARPONG,
FULGENCE MISHILI, JOYCE KINABO, AND JENNIFER C. COATES
Policies and programs often aim to improve the affordability of nutritious diets, but existing food price
indexes are based on observed quantities that may not meet nutritional goals. To measure changes in
the cost of reaching international standards of diet quality, we introduce a new cost of diet diversity in-
dex based on the lowest-cost way to include at least five different food groups as defined by the widely
used minimum dietary diversity for women (MDD-W) indicator and compare that to a Cost of
Nutrient Adequacy indicator for the lowest-cost way to meet estimated average requirements of
essential nutrients and dietary energy. We demonstrate application of both indexes using national av-
erage monthly prices from two very different sources: an agricultural market information system in
Ghana (2009–14) and the data used for national consumer price indexes in Tanzania (2011–15). We
find that the cost of diet diversity index for Ghana fluctuated seasonally and since mid-2010 rose about
10% per year faster than national inflation, due to rising relative prices for fruit, which also drove up
the cost of nutrient adequacy. In Tanzania there were much smaller changes in total daily costs, but
more adjustment in the mix of food groups used for the least-cost diet. These methods can show where
and when nutritious diets are increasingly (un)affordable, and which nutritional criteria account for
the change. These results are based on monthly national average prices, but the method is generaliz-
able to other contexts for monitoring, evaluation, and assessment of changing food environments.
Key words: Food prices, diet quality, diet diversity, nutrient adequacy, CPI.
JEL codes: I15, Q11, Q18.
Price indexes for traded food commodities Organization (FAO 2018), while local whole-
are widely reported by international agencies sale and retail prices are collected and used
such as the Food and Agricultural in almost all countries to monitor producer
prices, market conditions, overall inflation
This article was invited by the President of the Agricultural &
Applied Economics Association for presentation at the 2018 an-
nual meeting of the Allied Social Sciences Association, after which numerous workshop and seminar participants for their collabora-
it was subjected to an expedited peer-review process. tion and comments on this work. Data collection and analysis for
this project occurred under a project entitled Indicators of
William A. Masters is a professor, Yan Bai is a doctoral candidate, Affordability of Nutritious Diets in Africa (IANDA), funded by
Anna Herforth is a consultant and Jennifer C. Coates is an associ- UKAid through the Department for International Development
ate professor, all in the Friedman School of Nutrition Science and (DFID) as part of its program on Innovative Methods and Metrics
Policy at Tufts University; Daniel Sarpong is Dean for the School for Agriculture-Nutrition Actions (IMMANA), with additional
of Agriculture and an associate professor in agricultural economics support for data analysis from the Feed the Future Policy Impact
at the University of Ghana; and Fulgence Mishili is a senior lec- Study Consortium as a subaward from Rutgers University under
turer in agricultural economics and Joyce Kinabo is a professor of USDA Cooperative Agreement TA-CA-15-008, and the Feed the
human nutrition, both at Sokoine University of Agriculture in Future Innovation Lab for Nutrition under USAID grant contract
Tanzania. This paper was presented in a session on Agricultural AID-OAA-L-10-00006. We are especially grateful for a successor
Production, Diets and Health at the annual meetings of the ASSA, project on Changing Access to Nutritious Diets in Africa and
5 January 2018. We thank Andrew Dillon and other participants in South Asia (CANDASA) to extend this work, with funding from
that session as well as the AJAE editor, Travis Lybbert, and two UKAid and the Bill & Melinda Gates Foundation (OPP1182628).
anonymous reviewers for very valuable feedback on this paper, Model code and data for replication of results is available on that
and thank John Nortey, Rebecca Heidkamp, Zachary Gersten and project’s website at http://sites.tufts.edu/candasa.
Amer. J. Agr. Econ. 100(5): 1285–1301; doi: 10.1093/ajae/aay059
Published online August 13, 2018
VC The Author(s) 2018. Published by Oxford University Press on behalf of the Agricultural and Applied Economics
Association. This is an Open Access article distributed under the terms of the Creative Commons Attribution License
(http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any
medium, provided the original work is properly cited.
Downloaded from https://academic.oup.com/ajae/article-abstract/100/5/1285/5073250 by University of Ghana. Balme Library user on 24 July 2019
1286 October 2018 Amer. J. Agr. Econ.
and living standards (World Bank 2017a, assistance programs meet specific needs of
2017b). Formulas to aggregate individual items children and other vulnerable groups, as in
into price indexes were first introduced more the Cost-of-the-Diet approach developed by
than 300 years ago (Diewert 1993), with contin- Save the Children UK and others (Chastre
ued changes needed to reflect what and how et al. 2007; Deptford et al. 2017; Akhter et al.
goods and services are consumed (Diewert, 2018), and Optifood developed by the London
Greelees, and Hulten 2010; Rippy 2014). School of Hygiene and Tropical Medicine and
The purpose of most price indexes is to others (Optifood 2012; Vossenaar et al. 2017).
capture changes in the cost of what is actually Our aim in this paper is to extend the litera-
bought and sold, which can vary greatly in ture on the cost of nutritious diets to diversity
nutritional quality over time and across among food groups. Consuming foods from a
groups (Beatty, Lin, and Smith 2014; variety of different categories is often seen as
Clements and Si 2018). To make nutritious desirable for reasons beyond nutrient ade-
diets more affordable, policies and programs quacy, leading nutritionists to standardize diet
may aim to lower the relative cost of more diversity measurement by grouping foods in
nutritious foods, and sometimes also raise the terms of various functional characteristics. The
cost of less healthy items. The aim of this pa- specific criteria we use in this paper are from
per is to develop improved indexes for the the Minimum Diet Diversity for Women
cost of a nutritious diet relative to other pri- (MDD-W) indicator for women of reproduc-
ces in the African context, where healthier tive age (FAO and FHI360 2016; Martin-
foods such as dairy, eggs, fruits and vegeta- Prevel et al. 2017). MDD-W is defined as con-
bles vary greatly in price (Green et al. 2013; suming foods from at least five out of ten spe-
Harttgen, Klasen, and Rischke 2016). cific food groups during the previous day or
The oldest and most widely used approach night. This has been linked to nutrient ade-
to measuring the cost of healthy diets is the quacy in several low-income countries
cost of nutrient adequacy. Soon after the dis- (Arimond et al. 2010) and may confer addi-
covery of essential nutrients, Stigler (1945) tional health benefits associated with phyto-
pioneered the development of linear program- chemicals and other diet qualities in addition
ming methods for calculating how much of to nutrients (Shiraseb et al. 2016). Different
each food would be needed to meet recom- functional groups and thresholds could be used
mended intake of each required nutrient at for other populations at risk of malnutrition.
lowest total cost. Allen (2017) uses this kind For example, the standard international indica-
of price index for poverty measurement, and tor for dietary diversity in children aged 6–
others use these least-cost diets to track the 23 months is whether they consumed at least
cost of nutrients over time (O’Brien-Place one item from at least four out of seven spe-
and Tomek 1983; Håkansson 2015; Omiat and cific food groups in the previous day or night
Shively 2017), make comparisons across coun- (World Health Organization [WHO] and
tries (Chastre et al. 2007) or compare actual UNICEF 2007; UNICEF 2016). Dietary diver-
choices to least-cost diets within a country sity based on the number of food groups con-
(Jensen and Miller 2010; Maillot et al. 2017). sumed in the past 24 hours is operationally
Least-cost diets are often used to make nutri- useful for policy analysis and program manage-
tional recommendations for low-income con- ment, since it can be measured quickly using a
sumers. At the United States Department of list-based method, whereas the volume of food
Agriculture (USDA), the “Minimum-Cost consumed and its nutrient composition are
Food Plan” proposed for people facing ex- much more difficult to quantify and analyze.
treme poverty during the depression of the Designing a food price index around this crite-
1930s (Cofer et al. 1962) evolved with the use rion allows us to determine whether including
of linear programming into the Thrifty Food diverse foods in the diet is increasingly (un)af-
Plan (TFP) to calculate and justify the amount fordable for consumers at each time and place,
of money provided in food stamps and supple- to reach a minimum number of groups or to in-
mental nutrition assistance for low-income clude at least one item from every food group.
Americans (USDA 2017). The same method
is used internationally, for example to make
recommendations in Denmark (Parlesak et al. Methods
2016) and the Netherlands (Gerdessen and
De Vries 2015). One of the most important To track changes in the cost of nutritious
uses for least-cost diets is to help nutrition diets with broad relevance for the adult
Downloaded from https://academic.oup.com/ajae/article-abstract/100/5/1285/5073250 by University of Ghana. Balme Library user on 24 July 2019
Masters et al. Measuring the Affordability of Nutritious Diets in Africa 1287
population, we compute a price index defined consumed above one serving of each food, so
around the MDD-W and compare that to the prices and hence the overall index are stan-
corresponding cost of nutrient adequacy, us- dardized to cost per kcal. Additional informa-
ing monthly national average food prices in tion in the supplementary online material is
Ghana and Tanzania. We refer to the two provided on costs per gram. Also by defini-
measures as the cost of dietary diversity tion only the least-cost food within each
(CoDD), defined as the least-cost foods group is included, so the foods included in
needed to meet the MDD-W, and the cost of CoDD are not necessarily a positive descrip-
nutrient adequacy (CoNA), defined as the tion of what people actually consume or a
least-cost foods needed to meet average nu- normative prescription for what they should
trient requirements. Both are computed rela- consume. Instead, CoDD1 provides a lower
tive to all other prices in the local economy bound on the cost of including the fifth group
and converted to constant US dollars at to just meet the MDD-W threshold, while
purchasing-power parity (PPP) exchange CoDD2 provides a lower bound on the cost
rates. This provides comparable inflation- of acquiring some of each MDD-W food
adjusted price indexes, measuring the cost of group, thereby tracking changes in access to
reaching these two nutritional standards rela- foods needed to meet that nutritional
tive to all prices in the economy. Some foods standard.
appear in our nutritional price index also ap- As a benchmark for comparison we use the
pear in the PPP price level but with different same data to compute the cost of nutrient ad-
weights. Overall inflation is based on average equacy (CoNA), defined as the minimum
expenditure in each country as computed by cost of foods that meet all known require-
the International Comparisons Project ments for essential nutrients and dietary en-
(World Bank 2017a), while our indexes are ergy requirements for an adult woman of
based on each food’s contribution to meeting reproductive age. While we focus on women
international nutrition standards. of reproductive age for both CoDD and
Our price index for the cost of dietary di- CoNA, CoNA can be computed for other age
versity, CoDD, is defined as the least expen- or population groups, such as young children,
sive way of acquiring some food from each simply by using energy and nutrient require-
food group needed to reach the MDD-W. To ments specific to those groups. CoNA can be
aggregate over groups, we provide two dis- written formally as:
tinct measures: a simple CoDD1 counts only X
the least-cost food in the fifth least expensive ð3Þ CoNA : minimize C ¼ pi  qi
food group, while a broader CoDD2 counts i
the average of the least-cost food in all food Subject to:
groups. CoDD1 reflects a narrow version of
the MDD-W defined so that dietary diversity X
can be achieved just by reaching the fifth ð4Þ aij  qi  EAR ðj ¼ 1;2;3; : : : ; nÞi
group, while CoDD2 reflects a broader ver- X
sion in which consumers include all food ð5Þ qie  qi ¼ Ei
groups with equal frequency. The CoDD con-
cept is based on rank order optimization ð6Þ q1  0; q2  0; : : : ; qi  0
based on food prices within groups, defined
as: Here the quantity of the jth nutrient in
food i is denoted aij, which multiplied by its
ð1Þ CoDD1 ¼ min5fminf g quantity consumed (qi) must meet the popu-pi1 ; lation’s estimated average requirement
minfpi2g; : : : ; minfpimgg (EAR) for nutrient j, at lowest total cost
ð Þ ¼ f f g given all prices (pi) within the further con-2 CoDD2 ave min pi1 ; straint of overall energy balance (E) which
minfpi2g; : : : ; minfpimgg for convenience we set at 2,000 kcal/day.
There are twenty-one known essential
where min5 denotes the fifth lowest of all m nutrients, but for nutritional adequacy we
food groups, and pij is the price of item i in drop vitamin D and cholesterols, which can
the jth food group. By definition, the MDD- be synthesized in human bodies, and iodine
W indicator and hence CoDD price indexes and molybdenum due to lack of data in the
make no reference to the quantities food composition databases, leaving n¼ 17
Downloaded from https://academic.oup.com/ajae/article-abstract/100/5/1285/5073250 by University of Ghana. Balme Library user on 24 July 2019
1288 October 2018 Amer. J. Agr. Econ.
nutrient constraints plus a constraint for en- for nutrients may differ, we construct a semi-
ergy balance. This computation provides a elasticity denoted SP’ as increment in cost of
lower bound on the cost of meeting the the CoNA diet when the constraint is in-
EARs, allowing us to track changes in the creased by 1%, expressed as:
cost of limiting nutrients much as the CoDD
tracks changes in the cost of limiting foods. @C0
For both CoDD and CoNA we report ð8Þ SPj ¼ %DEARþ
which foods would be needed to meet each j
nutritional target at lowest cost, thereby 0 P ’ ’
tracking changes in access to that interna- The sum of SPj ( j SPj or SP ) of all sev-
tional standard. By defining “access” to mean enteen nutrients and dietary energy equals to
a lower bound on total cost, these price in- the change of CoNA when all nutritional and
dexes deliberately differ from what any group energy constraints are increased by 1% to-
might actually consume (for which we would gether. For ease of comparison with CoNA
use a consumption price index), or should itself we report SP’ multiplied by 100, which
consume (in the sense of a recommended we refer to as the shadow price contribution
diet). Actual diets may exceed or fall short of (SPC) of nutrient j or dietary energy:
any given nutritional standard, and methods 0
designed to make dietary recommendations ð9Þ SPCj ¼ SPj  100
include additional stipulations to obtain a lo-
cally acceptable, “normal” diet (Chastre et al. Similarly, we further calculated the shadow
2007; Deptford et al. 2017; Cost of Nutritious price elasticity (SPE) of nutrient j defined as
Diets Consortium 2018). For our purposes, the percentage change of the cost of the
we avoid specifying local eating habits and CoNA diet package evaluated at the optimal
cultural norms so as to compare food prices basis in response of 1% increase in EAR of
only with respect to international standards nutrient j:
for nutritional content. The foods whose pri-
ces are included in CoDD and CoNA are the %DC
lowest cost way to meet those standards at ð10Þ SPEj ¼ %DEARþ
each location, which may point to foods that j
are currently consumed in small quantities
The SPE is useful to identify the limiting
but could play a larger role in local diets if cu-
nutrients for which the level of EAR contrib-
linary practices were to change in response to
utes the most to CoNA at each time and
relative prices. Apart from those two price in-
place. It measures the change in total cost as-
dices presented here, parallel work is under
sociated with a marginal change in each nutri-
way to construct nutritionally weighted con-
ent requirement, thereby revealing the
sumer price indexes (nCPI) that would reflect
degree to which that particular requirement
nonmarket (dis)utilities from the foods actu-
accounts for differences in the cost of acquir-
ally consumed, and to construct globally rele-
ing all essential nutrients.
vant cost of a recommended diet (CoRD)
Calculations for all equations were com-
indexes that would reflect normative dietary
pleted in R and resulting index values
guidelines published by national or interna-
exported to Stata or Excel for visualization
tional agencies (Herforth 2017).
purposes, with model code and data for repli-
The focus of CoNA is the cost of nutrients,
cation posted online at the project website
which is reflected in their shadow prices (SP)
referenced in this paper’s acknowledgements.
defined as the cost increase associated with
increasing each constraint by one unit:
Data
ð Þ ¼ @C

7 SPj
@EARþj Our empirical application draws on four
main data sources. Food price data are na-
where C* denotes the (minimum) cost of tional average monthly food prices in Ghana
the CoNA diet. SPj is the SP of nutrient j between March 2009 and December 2014,
(or daily dietary energy), and EARþj refers to and in Tanzania between January 2011 and
one unit increase in EAR of nutrient j (or December 2015. These were collected by na-
daily dietary energy). Since units of measure tional authorities and cover a total of 34
Downloaded from https://academic.oup.com/ajae/article-abstract/100/5/1285/5073250 by University of Ghana. Balme Library user on 24 July 2019
Masters et al. Measuring the Affordability of Nutritious Diets in Africa 1289
Table 1. Descriptive Statistics for Monthly Food Prices per 1,000 kcal – Ghana (2011$)
Food Groups No. Foodstuffs Obs. Mean Std. Dev. CV Min Max
Grains, white roots and 1 Cassava 70 0.33 0.07 0.20 0.23 0.48
tubers, and plantains 2 Cocoyam 70 1.07 0.24 0.23 0.71 1.62
3 Kokonte 70 0.38 0.06 0.17 0.27 0.54
4 Garri 70 0.44 0.07 0.17 0.34 0.72
5 Rice (imported) 70 0.73 0.12 0.16 0.60 1.09
6 Rice (local) 70 0.52 0.06 0.12 0.42 0.75
7 Maize 70 0.26 0.05 0.18 0.19 0.40
8 Millet 70 0.39 0.05 0.13 0.31 0.51
9 Paddy Rice 56 0.40 0.13 0.32 0.24 0.86
10 Plantains 70 1.47 0.49 0.33 0.91 3.38
11 Sorghum 70 0.37 0.04 0.11 0.29 0.47
12 Yam 70 1.04 0.17 0.16 0.76 1.48
Pulses 13 Cowpea 70 0.61 0.10 0.17 0.43 0.85
14 Soybeans 70 0.29 0.07 0.24 0.13 0.47
Nuts and seeds 15 Groundnuts (shelled) 70 0.58 0.11 0.19 0.40 0.79
Meat, poultry, and fish 16 Anchovies 70 4.83 1.04 0.22 2.43 8.92
17 Tilapia (dried) 70 2.53 0.61 0.24 1.03 4.32
18 Herring (smoked) 70 1.99 0.45 0.22 1.27 3.45
Eggs 19 Eggs 70 6.23 0.44 0.07 5.22 7.58
Vitamin A-rich vegetables 20 Mangoes 70 1.41 0.51 0.36 0.64 2.94
and fruits
Other vegetables 21 Eggplant 70 9.16 2.37 0.26 4.78 16.55
22 Onions (large) 70 8.95 2.90 0.32 4.20 14.51
23 Tomatoes 70 20.77 6.88 0.33 10.09 39.91
Other fruits 24 Bananas 70 1.90 0.37 0.20 1.15 2.84
25 Oranges 70 2.94 0.90 0.31 1.20 6.72
26 Pineapples 70 2.94 0.32 0.11 2.29 3.87
Note: Authors’ calculations, from Ghana Ministry of Food and Agriculture file data. Two food groups in the MDD-W are not represented in this data set:
Dairy, and Dark Green Leafy Vegetables. Data were imputed by carry-over from the previous month to fill missing observations for soybeans (Feb 2010)
and mango (Aug, Sep, and Oct 2009; Feb 2011; Sep and Oct 2013). Kokonte and garri are forms of processed cassava.
distinct foods in Ghana and 71 in Tanzania. extremes, as mangoes in Ghana generally ma-
Prices for each item are unweighted averages ture between May and August, with some va-
over a variety of retail markets, covering all rieties in southern Ghana also maturing
ten regions of Ghana and all twenty-one between December and February (MoFA
regions of mainland Tanzania. Primary data 2017).
collection was conducted by the Ministry of To compute the price indexes, the price of
Food and Agriculture (MoFA) in Ghana for each food was converted from reported units,
their market information system, and by the such as price per dozen eggs, to cost per unit
National Bureau of Statistics (NBS) in of weight and/or of dietary energy of the edi-
Tanzania for the purpose of inflation moni- ble portion, and then converted to a common
toring. In this paper we deliberately use data currency and adjusted for inflation by
with different institutional origins to show purchasing-power-parity (PPP) conversion
the range of applicability for these indexes, factor provided by the World Bank (2017a).
recognizing that differences between coun- We excluded most processed foods and clas-
tries also reflect differences in data-collection sified foods into one of ten mutually exclusive
methods. There were no missing values in the food groups based on the FAO and FHI360
Tanzania data, but for Ghana there are miss- (2016) guidelines for calculating MDD-W:
ing observations for soybeans (Feb 2010) and (1) Grains, white roots and tubers, and plan-
mango (Aug, Sep, and Oct 2009; Feb 2011; tains, (2) pulses, (3) nuts and seeds, (4) dairy,
Sep and Oct 2013). To complete the data set (5) meat, poultry and fish, (6) eggs, (7) dark
for results shown here we impute prices by green leafy vegetables, (8) vitamin A-rich
carry-over from the previous month. This fruits and vegetables, (9) other vegetables,
method is unlikely to truncate seasonal and (10) other fruits. Additional foods that
Downloaded from https://academic.oup.com/ajae/article-abstract/100/5/1285/5073250 by University of Ghana. Balme Library user on 24 July 2019
1290 October 2018 Amer. J. Agr. Econ.
Table 2. Descriptive Statistics for Monthly Food Prices per 1,000 kcal–Tanzania (2011$)
Food Group No. Foodstuff Obs. Mean Std. Dev. CV Min Max
Grains, white roots and 1 Cassava (dried flour) 60 0.60 0.07 0.11 0.48 0.79
tubers, and plantains 2 Cassava (fresh) 60 0.77 0.07 0.09 0.60 0.90
3 Plantain 60 1.64 0.09 0.05 1.45 1.90
4 Finger millet 60 0.68 0.11 0.17 0.50 0.87
5 Maize flour 60 0.47 0.06 0.12 0.37 0.63
6 Potatoes – round 60 2.25 0.13 0.06 1.97 2.63
7 Rice 60 0.74 0.12 0.16 0.57 0.98
8 Sweet potatoes 60 1.70 0.14 0.08 1.46 1.97
9 Wheat flour 60 0.62 0.04 0.06 0.56 0.71
10 Maize (white) 60 0.31 0.04 0.12 0.24 0.41
Pulses 11 Soybeans 60 0.65 0.03 0.04 0.59 0.70
12 Lentils 60 1.28 0.12 0.09 1.08 1.48
13 Beans (red) 60 0.78 0.04 0.05 0.72 0.87
Nuts and seeds 14 Groundnuts 60 0.66 0.05 0.08 0.58 0.78
Dairy 15 Milk (fresh) 60 2.89 0.16 0.05 2.38 3.07
16 Milk (powdered) 60 7.99 0.38 0.05 7.02 8.72
Meat, poultry, and fish 17 Beef sausage 60 4.32 0.08 0.02 4.18 4.54
18 Beef with bones 60 3.92 0.19 0.05 3.47 4.43
19 Beef without bones 60 1.11 0.04 0.04 1.01 1.26
20 Dried sardines 60 5.99 0.46 0.08 5.12 6.91
21 Goat meat 60 9.51 0.38 0.04 8.37 10.19
22 Chicken (live, industrial) 60 6.57 0.31 0.05 5.6 6.99
23 Pork meat 60 3.17 0.28 0.09 2.45 3.63
24 Chicken (live, traditional) 60 11.9 0.79 0.07 9.94 13.26
Eggs 25 Eggs (layers) 60 8.42 0.28 0.03 7.89 8.88
26 Eggs (traditional) 60 11.81 0.69 0.06 10.3 12.66
Dark green leafy vegetables 27 Amaranth (mchicha) 60 5.74 0.57 0.10 4.85 6.81
Vitamin A-rich 28 Carrots 60 7.05 0.69 0.10 6.01 9.08
vegetables and fruits 29 Mangoes 60 4.46 0.63 0.14 2.97 6.06
30 Papaya 60 5.63 0.50 0.09 4.71 6.64
Other vegetables 31 Tomatoes (bitter) 60 8.86 0.46 0.05 7.85 10.72
32 Eggplant 60 9.44 0.49 0.05 8.47 10.83
33 Cabbage 60 2.80 0.27 0.10 2.30 3.48
34 Green peas 60 24.78 1.74 0.07 20.72 28.40
35 Green bell pepper 60 16.46 0.92 0.06 14.78 19.16
36 Okra (ladies fingers) 60 11.28 0.75 0.07 9.97 13.25
37 Onions 60 6.43 0.77 0.12 5.21 8.86
38 Tomatoes (red) 60 10.44 1.19 0.11 8.36 13.53
Other fruits 39 Apples (imported) 60 19.58 1.62 0.08 15.85 23.62
40 Avocado 60 1.91 0.12 0.06 1.67 2.18
41 Coconut (mature) 60 5.52 0.51 0.09 4.78 6.85
42 Lemons 60 11.75 2.03 0.17 8.26 17.99
43 Limes 60 15.62 2.87 0.18 12.00 23.57
44 Oranges 60 4.43 0.46 0.10 3.47 5.63
45 Pineapples 60 6.66 0.65 0.10 5.54 7.98
46 Sweet banana 60 3.35 0.28 0.08 2.71 3.91
Note: Authors’ calculations, from Tanzania Bureau of Statistics file data.
people might consume are not included in the forty-six foods from all ten groups. The miss-
MDD-W calculation, notably oils and fats, ing food groups in Ghana are dairy and dark
sweets and other foods, beverages other than green leafy vegetables. We use these data to
dairy, condiments and seasonings. The avail- highlight that data gaps are often present
able price data for Ghana cover twenty-six in food price data monitoring systems,
foods from eight of the ten MDD-W food where staple crops are the focus and nutri-
groups, and price data for Tanzania cover tionally important foods may be missing.
Downloaded from https://academic.oup.com/ajae/article-abstract/100/5/1285/5073250 by University of Ghana. Balme Library user on 24 July 2019
Masters et al. Measuring the Affordability of Nutritious Diets in Africa 1291
2009m1 2009m7 2010m1 2010m7 2011m1 2011m7 2012m1 2012m7 2013m1 2013m7 2014m1 2014m7
Year/Month
CoDD1 index CoDD2 index Maize Cassava (dried) Cassava (fresh)
Soybeans Groundnuts Mangoes Banana Oranges
Tilapia (dried) Herrings (smoked) Eggs Eggplant (garden eggs) Onions (large)
Figure 1. Cost of diet diversity in Ghana, March 2009–December 2014.
Note: Foods shown are the least-cost item in their food group, as defined by the minimum dietary diversity for women (MDD-W) indicator, ranked in cost
per unit of dietary energy. CoDD1 is the cost of reaching the fifth group, and CoDD2 is the cost of including all groups. Groups in ascending order of usual
cost are starchy staples (maize and cassava), pulses (soybeans), nuts/seeds (groundnuts), vitamin-A rich fruits and vegetables (mangoes), other fruit (banana
and oranges), meat and fish (tilapia and herring), eggs, and other vegetables (eggplant and onion).
The MDD-W offers a quick way to assess defined as the average daily nutrient intake
whether monitoring systems cover sufficient level estimated to meet requirements at least
diversity to estimate the cost of nutritious half of the healthy individuals in a group, is
diets: if there are no or few items represent- the primary reference point for assessing the
ing each of the ten MDD-W categories, then adequacy of estimated nutrient intakes of
those data gaps should be corrected. Based groups and is a tool for planning intakes for
on these analyses, we collaborated with the groups (Institute of Medicine 2006). A de-
Ghana MoFA to add nutritious food items to tailed table with energy and nutrients criteria
their food price monitoring system (Nortey is presented in online supplementary material
2017). By definition, cooking oil is not in- table A3.
cluded in the MDD-W or CoDD, but we do
include it as a source of dietary energy for
CoNA. Results
Additional data required for the calcula-
tion of CoNA include the nutrient composi- Descriptive statistics for prices per unit of di-
tion and edible portions of each food as etary energy are summarized in tables 1
purchased, obtained from the two standard and 2. The underlying descriptive statistics
sources: FAO’s West African Food for prices per unit of weight are provided in
Composition Table (Stadlmayr 2012), com- the online supplementary material, as tables
plemented by the U.S. National Nutrient A3 and A4. For Ghana, we have a total of 70
Database for Standard Reference (USDA monthly observations from March 2009 to
2013). Detailed food lists with nutrients com- December 2014 for twenty-five items, and
positions for both countries are presented in fifty-six monthly observations from May 2010
online supplementary material tables A4 and to December 2014 for paddy rice. Of these,
A5. Nutrient requirements are obtained from twelve food items are in the starchy staple
the estimated average requirements (EARs) group, reflecting the strong focus of data col-
for adult women from nineteen to thirty years lection efforts on that category. The average
old, as specified in dietary reference intakes price of each item per 1,000 kcal ranges
(DRIs) developed by the U.S. Institute of widely, from $0.26 for maize to $20.77 for to-
Medicine of the National Academies. EAR, matoes, while prices per kg range from $0.53
Cost per 1,000 kcal in 2011 intl $ (log scale)
.2 1 3 10
Downloaded from https://academic.oup.com/ajae/article-abstract/100/5/1285/5073250 by University of Ghana. Balme Library user on 24 July 2019
1292 October 2018 Amer. J. Agr. Econ.
2011m1 2011m7 2012m1 2012m7 2013m1 2013m7 2014m1 2014m7 2015m1 2015m7 2016m1
Year/Month
CoDD1 index CoDD2 index Maize (white) Soybeans Groundnuts
Beef Avocado Cabbages Milk Mangoes
Papaya Amaranth leaves Eggs
Figure 2. Cost of diet diversity in Tanzania, January 2011–December 2015.
Note: Foods shown are the least-cost item in their food group that month, as defined by the minimum dietary diversity for women (MDD-W) indicator. Items
are ranked in cost per unit of dietary energy. CoDD1 is the cost of reaching the fifth group, and CoDD2 is the cost of including all groups. Groups in ascend-
ing order of usual cost are a starchy staple (maize), pulses (soybeans), nuts/seeds (groundnuts), meat or fish (beef for stew), other fruit (avocado), other vege-
table (cabbage), dairy (milk), vitamin-A rich fruit (mangoes or papaya), green leafy vegetables (amaranth), and eggs.
200
175 Ghana - CoDD1
Ghana - CoDD2
150 Tanzania-CoDD2
Tanzania-CoDD1
125
100
75
50
Figure 3. Cost of diet diversity indexes for Ghana and Tanzania, 2009–2015 (January
20115 100).
Source: Index values shown represent changes in the cost of meeting energy needs from diverse food groups, when using the lowest-priced food in each food
group as defined for the minimum dietary diversity for women (MDD-W) indicator. CoDD1 shows the cost of reaching five food groups, and CoDD2 shows
the average cost of including all available groups. Which foods and food groups are included in this minimally diverse diet varies over time and space. Data
shown here are relative to Jan. 2011 prices in real USD/kcal.
for cassava to $8.90 for eggs shown in online January 2011 to December 2015 for forty-six
supplementary material table A1. The volatil- items spanning ten food groups as the final
ity of food prices over time, as represented by data base for index calculation. Starchy sta-
coefficient of variation (CV), varies widely ples group, as the largest food group in terms
from 0.07 for eggs to 0.36 for mangoes. of the number of food items, contains 10
For Tanzania, we have a total of sixty items. Average prices per 1,000 kcal range
monthly observations over five years from from $0.31 for white maize to $24.78 for
Index value (Jan. 2011=100) 
per unit of dietary energy (kcal) Cost per 1,000 kcal in 2011 intl $ (log scale)
.2 1 3 10
Downloaded from https://academic.oup.com/ajae/article-abstract/100/5/1285/5073250 by University of Ghana. Balme Library user on 24 July 2019
Masters et al. Measuring the Affordability of Nutritious Diets in Africa 1293
green peas, and prices per kg range from
$1.11 for white maize to $39.56 for powered
milk. The volatility of prices ranges from a
CV of 0.02 for beef sausage and goat meat to
0.18 for limes.
Turning to the CoDD indexes, figure 1
presents results for Ghana, showing the price
per unit of dietary energy for the lowest cost
item in each food group. The lowest-cost
foods are usually one of the starchy staples
(either maize or cassava), but for several
months in 2009 and early 2010 the lowest-cost
calorie source was actually from the pulse
group (soybeans). The third least expensive
source is from the nut group (groundnuts),
followed by vitamin A-rich vegetables and
fruits (mangoes), and other fruits (bananas).
Occasionally, some form of fish (salted dried
tilapia or smoked herrings) becomes the fifth
group. The cost of reaching the MDD-W is
shown by the solid line tracing the price of in-
cluding that fifth group (CoDD1). An alter-
native measure showing the average cost of
including any group (CoDD2), shown by the
dashed line, is higher due to inclusion of the
most expensive food groups. A similar analy-
sis in terms of cost per unit of weight, shown
in the online supplementary material figure
A1, yields qualitatively similar results, except
that the “other vegetable” group (repre-
sented here by eggplants and onions)
becomes cheaper than groundnuts due to its
higher moisture content.
Results for CoDD in Tanzania are pre-
sented in figure 2, showing that the lowest-
cost food group per unit of dietary energy is
always the starchy staple (maize), with a
pulse (soybean) and a nut (groundnuts) alter-
nating as the second and third least-costly
food group, followed by a meat (beef) and a
food from the “other fruit” group (avocado).
This figure reveals much more stability
among the lower-cost food groups than
among these foods in Ghana or relative to
more expensive food groups in Tanzania.
Such differences could reflect the type of
market at which food prices are collected, as
NBS in Tanzania aims to collect price data
for inflation monitoring from the same sellers
every time primarily in towns and cities,
whereas MoFA in Ghana aims to collect
price data for market information purposes
from different sellers every time, in a wider
variety and greater number of locations. In
Tanzania, the relative cost of foods by unit of
weight as shown in the online supplementary
material figure A2 differs greatly from cost
Table 3. Foods Selected for CoNA Diet Plans in Ghana, Mar 2009–Dec 2014
2009–2014 2009 2010 2011 2012 2013 2014
Food Item Mean %Selected Mean %Selected Mean %Selected Mean %Selected Mean %Selected Mean %Selected Mean %Selected
Cassava 21 11% 18 10%   47 25% 63 33%    
Maize 50 69% 14 20% 48 67% 55 75% 36 50% 66 92% 74 100%
Mangoes 900 100% 910 100% 904 100% 902 100% 905 100% 881 100% 899 100%
Paddy Rice 14 49%   6 25% 18 67% 13 50% 15 42% 27 100%
Palm Oil 4 51% 7 100% 6 75% 3 33% 4 50% 5 58%  
Plantain 3 1%         19 8%  
Smoked Herrings 15 100% 15 100% 15 100% 15 100% 15 100% 15 100% 15 100%
Soybeans 256 100% 289 100% 267 100% 242 100% 252 100% 246 100% 243 100%
Note: Data shown are mean intake (g/day) and intake frequency (percent of days) for lowest-cost diets that reach the estimated average requirement (EAR) of essential nutrients for an adult woman of 55 kg at an energy level of
2,000 kcal/day.
Downloaded from https://academic.oup.com/ajae/article-abstract/100/5/1285/5073250 by University of Ghana. Balme Library user on 24 July 2019
1294 October 2018 Amer. J. Agr. Econ.
2.50
2009 2010 2011 2012 2013 2014
2.00
Vit. A-rich fruits &
1.50 veg.
Other oils & fats
1.00 Meat, poultry & fish
Pulses
0.50
Starchy staples
0.00
Figure 4. Cost of nutrient adequacy by food group in Ghana, Mar 2009–Dec 2014.
Note: Data shown are total cost in each month of the foods needed for lowest cost of nutrient adequacy (CoNA), for an adult woman of 55kg at a dietary en-
ergy level of 2,000 kcal/day.
per unit of energy, due to the inclusion of enter with mean intakes of 900 and 256 g/day
foods with high moisture content notably respectively, as they are the principal sources
cabbage, amaranth leaves and milk. of limiting nutrients which are more costly to
To compare changes over time in the mar- obtain from other sources in the Ghanaian
ginal cost of obtaining any amount of dietary context. Such a high level of consumption for
energy from the fifth food group (CoDD1) or these two foods is not realistic or recom-
the average of all food groups (CoDD2), it is mended, but does reveal the degree to which
convenient to standardize costs as index num- the nutrient profile of mango and soybean fills
bers. Results shown in figure 3 reveal differen- gaps left by other foods listed in table 3 below.
ces relative to the base period of January 2011, As shown in figure 4, the CoNA index for
after which CoDD1 rose sharply with wide Ghana more than doubled from USD 0.78
swings in Ghana and changed much less in per day in March 2009 to USD 1.87 in
Tanzania. Using all food groups instead of just December 2014. We can link the foods that
the fifth, CoDD2 rose less than CoDD1 in account for this rise back to the food groups
Ghana but more in Tanzania, due to differen- used for CoDD, noting that mangoes from
ces in price trends among the most and least the vitamin A-rich fruits and vegetables
expensive foods. A similar chart based on cost group accounted for more than 60% of
per unit of weight is provided in our online CoNA on average. Soybeans from the pulses
supplementary material figure A3, revealing group contributed about 28% of CoNA on
qualitatively similar results in most months. average, while cassava from the starchy sta-
Focusing on our preferred cost of dietary di- ples group, and smoke herrings from the
versity measure in figure 3, CoDD1 per unit of flesh-foods group accounted for approxi-
dietary energy rose by about 50% from mately 6% and 4%, respectively. The remain-
January 2011 to late 2014, while in Tanzania, ing cost was palm oil, which is not included in
the price indices were relatively stable from CoDD and which contributed about 1.5% of
January 2011 to December 2015. CoNA before July 2013, then not selected for
We can compare the cost of dietary diver- least-cost diet packages thereafter.
sity by food group to the cost of nutrient ade- As shown in table 4, in Ghana a total of
quacy (CoNA) using each month’s solution five nutrients have limiting EARs, four of
to equations (3) – (6). For Ghana, a total of which were limiting nutrients in all months.
eight distinct food items are ever included in Vitamin A, as the most expensive nutrient,
those least-cost diets. Three of these foods has a shadow price elasticity (SPE) of 0.47,
(mangoes, soybeans and smoked herring) are meaning that CoNA increases by 0.47%
included every month. Mangoes and soybeans when the EAR for vitamin A increases by
US dollars per day (at 2011 PPP prices)
Downloaded from https://academic.oup.com/ajae/article-abstract/100/5/1285/5073250 by University of Ghana. Balme Library user on 24 July 2019
Masters et al. Measuring the Affordability of Nutritious Diets in Africa 1295
1%, that is, from 500 mcg to 505 mcg per day.
Dietary energy is still a very important con-
straint in Ghana with an average SPE of 0.34.
As shown in figure 5, the nutrients that are
most limiting for CoNA in Ghana are vitamin
A, followed by dietary energy, vitamin E, cal-
cium and vitamin B12.
For Tanzania, the CoNA solution to equa-
tions (3) – (6) spans sixty months from
January 2011 to December 2015. As shown in
table 5, a total of eight food items are ever se-
lected, of which two (white maize and mchi-
cha or amaranth leaves) are included in every
month with mean intakes of 255 and 197 g/d,
respectively.
Figure 6 reveals that the CoNA indicator
for Tanzania fluctuated much less than for
Ghana, rising gradually by about 25%, from
1.17 to 1.48 USD/day over these five years,
but there is sharp variation in the composi-
tion of this least-cost diet over time. Pulses
enter periodically when soy is relatively inex-
pensive, displacing both the lowest-cost
starchy staple (maize) and the lowest-cost
green leafy vegetable (mchicha), and cooking
oil is displaced by nuts and seeds (ground-
nuts). A dried fish (sardines) remains in this
least-cost diet at about 10% of its total cost
throughout the period.
In Tanzania, as shown in table 6 there
were in total of seven limiting nutrients, in-
cluding the same five limiting nutrients as
Ghana plus vitamin C and selenium. Dietary
energy, calcium, vitamin C, B12 and E were
limiting nutrients in all observations. Using
the SPE as a criterion, dietary energy was
the most constraining nutritional factor in
Tanzania, as a 1% increase in daily dietary
energy requirement from 2,000 to 2,020 kcal
would increase CoNA by 0.4%. The most
constraining individual nutrient was calcium
with an average SPE of 0.3, meaning an in-
crease in CoNA of 0.3% if calcium require-
ments rose from 800 mg to 808 mg. Vitamin
A is less costly in Tanzania than in Ghana,
with an SPE of only 0.027. As shown in
figure 7, energy became increasingly con-
straining until early 2013, and then calcium
became relatively more important until early
2015 when the relative cost of acquiring die-
tary energy rose again.
Discussion and Conclusions
This paper presents nutritional price indexes
to compare the relative cost of reaching
Table 4. Nutrient Requirements Contributing to CoNA in Ghana, Mar 2009–Dec 2014
2009–2014 2009 2010 2011 2012 2013 2014
Nutrient %EAR SPE %EAR SPE %EAR SPE %EAR SPE %EAR SPE %EAR SPE %EAR SPE
Always limiting nutrients
Energy 100% 0.344 100% 0.423 100% 0.344 100% 0.402 100% 0.391 100% 0.213 100% 0.302
Vitamin B12 100% 0.029 100% 0.032 100% 0.029 100% 0.030 100% 0.027 100% 0.027 100% 0.032
Vitamin A 100% 0.467 100% 0.420 100% 0.448 100% 0.407 100% 0.470 100% 0.548 100% 0.500
Vitamin E 100% 0.086 100% 0.109 100% 0.116 100% 0.082 100% 0.049 100% 0.107 100% 0.058
Sometimes limiting nutrients
Calcium 104% 0.074 114% 0.016 107% 0.063 100% 0.079 103% 0.062 100% 0.104 100% 0.109
Note: Data shown are mean fraction of the estimated average requirement for an adult woman of 55 kg at an energy level of 2,000 kcal/day consumed each day (%EAR). The mean shadow price elasticity (SPE) of each nutrient when it is lim-
iting. SPE is defined as the percentage change of CoNA if the EAR for that nutrient were increased by 1%.
Downloaded from https://academic.oup.com/ajae/article-abstract/100/5/1285/5073250 by University of Ghana. Balme Library user on 24 July 2019
1296 October 2018 Amer. J. Agr. Econ.
2.50
2009 2010 2011 2012 2013 2014
2.00
1.50 Vitamin A
Calcium
Vitamin E
1.00
Vitamin B12
Energy
0.50
0.00
Figure 5. Cost of nutrient adequacy by limiting nutrient in Ghana, Mar 2009–Dec 2014.
Note: Data shown are total cost in each month of the foods needed for lowest cost of nutrient adequacy (CoNA), for an adult woman of 55 kg at a dietary en-
ergy level of 2,000 kcal/day.
international nutrition standards at each time nutrient needs in both countries is also
and place. We introduce a cost of diet diver- heavily driven by daily energy requirements,
sity (CoDD) index, defined as the minimum with each 1% rise in energy intake leading to
cost of acquiring at least five out of ten spe- an 0.3–0.5% rise in the least-cost diet.
cific food groups, for comparison with the Comparing results for CoDD and CoNA
cost of nutrient adequacy (CoNA), which reveals the continued importance of year-
tracks the minimum cost of meeting esti- round access to basic staple foods for macro-
mated average requirements of energy, pro- nutrients, while identifying the other foods
tein and seventeen essential nutrients. These and specific micronutrients that limit access
indexes reveal temporal and spatial differen- to healthier diets.
ces in access to diverse diets and adequate The CoDD and CoNA indexes are
nutrients, helping to guide policies and pro- intended to track access and affordability of
grams aimed at improving their affordability foods required for a given nutritional stan-
in local markets. dard, which may be very different from what
Using national average monthly prices for is actually consumed. Actual diets often fall
Ghana from 2009 through 2014, we find that short of international standards for some
the cost of meeting the diet diversity standard nutrients, while exceeding minimal needs in
fluctuated seasonally and rose sharply from other dimensions, driven in part by relative
mid-2010 through 2014 at about 10% per prices among different foods. CoDD is a
year faster than inflation, due primarily to ris- unit-free measure to track changes in relative
ing relative prices for fruit. The cost of nutri- prices, while CoNA is a cost per day which
ent adequacy doubled over this period, due we can compare to income levels or actual
primarily to increased cost of foods needed to expenditure. In 2012, for example, the level
meet standards for vitamin A and also cal- of CoNA in both Ghana and Tanzania was
cium. Similar monthly data for Tanzania about $1.40 per person at 2011 PPP prices,
show an upward trend from 2011 to 2012 and while national average per-capita food ex-
then seasonal fluctuations through 2015, penditure in rural areas was estimated at
switching among different leguminous grains $1.73 in Tanzania and $2.99 in Ghana (IFPRI
and green leafy vegetables as the lowest-cost 2017). Low incomes make nutritious diets out
way to meet nutrient needs. In both Ghana of reach for many people, especially in
and Tanzania, vitamin B-12 needs lead to in- Tanzania, but even when incomes are higher
clusion of dried fish in CoNA indexes, even as in Ghana the relative cost of different
though it is not included for diet diversity foods will influence food choice. CoNA is
purposes in CoDD. The cost of meeting particularly useful for identifying foods such
US dollars per day (at 2011 PPP prices)
Downloaded from https://academic.oup.com/ajae/article-abstract/100/5/1285/5073250 by University of Ghana. Balme Library user on 24 July 2019
Masters et al. Measuring the Affordability of Nutritious Diets in Africa 1297
as soybeans that have recently become low-
cost sources of essential nutrients, and might
therefore play an increasing role in local diets
as culinary practices evolve.
Nutritional price indexes like CoDD and
CoNA can guide public investment, policies
and programs that make high-quality diets
more accessible year-round even in remote
areas, complementing farmers’ self-provision-
ing with interventions that lower the relative
cost of nutrient-dense foods on local markets.
Introducing these indexes could increase de-
mand for price data about a wider range of
nutritious foods at various places. This paper
uses two very different data sources: the
Ministry of Food and Agriculture (MoFA)
market information service in Ghana, and the
National Bureau of Statistics system for mon-
itoring inflation in Tanzania. The Ghana data
are intended to inform production and mar-
keting of traded crops, and historically omit-
ted two nutritionally important kinds of food:
dark green leafy vegetables and dairy.
Demand for data to construct CoDD and
CoNA has already led MoFA to expand their
market information service to a wider range
of foods (Nortey 2017), with preliminary
analysis of the new data by the World Food
Program revealing additional opportunities
to meet nutritional targets at lower cost be-
yond the results in this paper (WFP 2017).
Using CoDD and CoNA can also guide sta-
tistical agencies to collect price data at times
and places when nutritious diets are most out
of reach. For example, ongoing studies using
the Tanzania data identify regional differen-
ces in seasonality (Bai et al. 2018), revealing
where and when price fluctuations are most
important to measure and eventually address
with investments in market infrastructure tai-
lored to the specific foods needed for more
nutritious diets.
Our CoDD index aims to guide interven-
tions that target diet diversity, in this case to
reach the MDD-W threshold of at least five
food groups from a list of ten. This is useful
for settings where diversity as such is impor-
tant (Clements and Si 2018), across specific
food groups as in the U.S. Healthy Eating
Index used by Beatty, Lin, and Smith (2014)
for the United States. In developing coun-
tries, many agencies aim explicitly to increase
the proportion of women whose diets meet
the MDD-W threshold (e.g., Feed the Future
2018), and the CoDD price index can reveal
which foods drive the cost of reaching that
goal at each time and place. CoDD can also
Table 5. Foods Consumed to Construct CoNA Diet Plans in Tanzania, Jan 2011–Dec 2015
2011–2015 2011 2012 2013 2014 2015
Mean Pct. of Mean Pct. of Mean Pct. of Mean Pct. of Mean Pct. of Mean Pct. of
Food Item g/day months g/day months g/day months g/day months g/day months g/day months
Cabbages 25 80% 14 33% 24 67% 34 100% 26 100% 27 100%
Cassava flour 0.1 7%   1 33%      
Cooking oil 9.2 32% 29 100% 12 42%     4 17%
Dried sardines 14 100% 14 100% 14 100% 14 100% 14 100% 14 100%
Mchicha (amaranth) 197 100% 263 100% 203 100% 143 100% 188 100% 187 100%
Groundnuts 89 68%   74 58% 127 100% 131 100% 111 83%
Soybeans 92 78% 69 100% 91 92% 122 83% 85 58% 92 58%
White Maize 255 100% 369 100% 269 100% 183 100% 219 100% 232 100%
Note: Data shown are mean intake (g/day) and intake frequency (percent of days) for lowest-cost diets that reach the estimated average requirement (EAR) of essential nutrients for an adult woman of 55 kg at an energy level of
2,000 kcal/day.
Downloaded from https://academic.oup.com/ajae/article-abstract/100/5/1285/5073250 by University of Ghana. Balme Library user on 24 July 2019
1298 October 2018 Amer. J. Agr. Econ.
Figure 6. Cost of nutrient adequacy by food group in Tanzania, Jan 2011–Dec 2015.
Note: Data shown are total cost in each month of the foods needed for lowest cost of nutrient adequacy (CoNA), for an adult woman of 55 kg at a dietary en-
ergy level of 2,000 kcal/day.
Table 6. Nutrient Requirements Contributing to CoNA in Tanzania, Jan 2011–Dec 2015
2011–2015 2011 2012 2013 2014 2015
% % % % % %
Nutrient EAR SPE EAR SPE EAR SPE EAR SPE EAR SPE EAR SPE
Always limiting nutrients
Energy 100% 0.433 100% 0.415 100% 0.430 100% 0.498 100% 0.422 100% 0.401
Calcium 100% 0.296 100% 0.312 100% 0.266 100% 0.254 100% 0.318 100% 0.327
Vitamin E 100% 0.157 100% 0.179 100% 0.189 100% 0.127 100% 0.139 100% 0.150
Vitamin B12 100% 0.080 100% 0.077 100% 0.079 100% 0.075 100% 0.083 100% 0.031
Sometimes limiting nutrients
Vitamin A 100% 0.027 100% 0.011 100% 0.030 100% 0.031 100% 0.032 100% 0.031
Vitamin C 154% 0.006 198% 0.001 158% 0.003 119% 0.014 147% 0.006 147% 0.005
Folate 228% 0.001 151% 0.003 218% 0.004 278%  250%  243% 
Note: Data shown are mean fraction of the estimated average requirement for an adult woman of 55 kg at an energy level of 2,000 kcal/day consumed each
day (%EAR). The mean Shadow Price Elasticity (SPE) of each nutrient when it is limiting. SPE is defined as the percentage change of CoNA if the EAR for
that nutrient were increased by 1%.
inform where and when the cost of reaching data collection, but CoNA suggests that they
MDD-W is highest, to help target interven- often play an outsized role in the cost of nutri-
tions towards more universal access to ade- tious diets. Using CoNA can guide interven-
quate dietary diversity. tions toward the lowest-cost way to achieve
The CoNA index adds information about many aspects of food security targeted by
the quantities of each food needed to meet international agreements (FAO 1996) and re-
nutritional targets, identifying which veal which nutrients remain most difficult to
nutrients are most expensive and which foods obtain even after policies and programs at-
are most-effective at each time and place. tempt to expand food access. For those
Our results reveal the universal importance nutrients, fortification and supplementation
of a few staple foods to meet macronutrient are important options, as shown by an analy-
constraints in both Ghana and Tanzania, sis of our data regarding opportunities for sta-
even as a variety of other foods are also ple flour fortification in Ghana (WFP 2017).
needed to reach micronutrient standards at In summary, the index proposed here for
different locations and times of year. Fruits the cost of diet diversity, alongside traditional
and vegetables have not traditionally been measures for the cost of nutrient adequacy,
prioritized for either public investment or allow us to measure changes in the
Downloaded from https://academic.oup.com/ajae/article-abstract/100/5/1285/5073250 by University of Ghana. Balme Library user on 24 July 2019
Masters et al. Measuring the Affordability of Nutritious Diets in Africa 1299
1.60
2011 2012 2013 2014 2015
1.40 Folate
1.20 Vitamin C
1.00 Vitamin A
0.80 Vitamin B12
0.60
Vitamin E
0.40
Calcium
0.20
Energy
0.00
Figure 7. Cost of nutrient adequacy by limiting nutrient in Tanzania, Jan 2011–Dec 2015.
Note: Data shown are total cost in each month of the foods needed for lowest cost of nutrient adequacy (CoNA), for an adult woman of 55 kg at a dietary en-
ergy level of 2,000 kcal/day.
(un)affordability of healthier diets than those References
currently consumed. Monitoring changes in
these indexes can reveal the degree to which Akhter, N., N. Saville, B. Shrestha, D.S.
policy and program interventions improve ac- Manandhar, D. Osrin, A. Costello, and
cess to nutritious diets, focusing on the needs A. Seal. 2018. Change in Cost and
of low-income people who seek the least ex- Affordability of a Typical and
pensive foods at each time and place. One Nutritionally Adequate Diet among
key limitation of the work so far is that exist- Socio-Economic Groups in Rural Nepal
ing price monitoring systems often miss foods after the 2008 Food Price Crisis. Food
of nutritional importance for the poor, and Security 10 (3): 615–629.
even when prices are available their nutri- Allen, R.C. 2017. Absolute Poverty: When
tional content may not have been measured. Necessity Displaces Desire. American
Future research using available and new data Economic Review 107 (12): 3690–721.
can identify which of the functional forms dis- Arimond, Mary, Doris Wiesmann, Elodie
cussed here are most sensitive to differences Becquey, Alicia Carriquiry, Melissa C.
in local food environments, and most predic- Daniels, Megan Deitchler, Nadia Fanou-
tive of differences in nutrition outcomes. Fogny, et al. 2010. Simple Food Group
Formulation of the indexes could also spur Diversity Indicators Predict
improvements in data collection systems, to Micronutrient Adequacy of Women’s
include all locally available foods that might Diets in 5 Diverse, Resource-Poor
help meet nutritional standards at prices that Settings. The Journal of Nutrition 140
accurately reflect the cost of acquisition for (11): 2059S–69S.
local households. More complete data collec- Bai, Y., E. Naumova, and W.A. Masters.
tion would, in turn, improve the value of new 2018. Seasonal Differences in the
indexes, to analyze how specific policies and Affordability of Nutritious Diets in
programs alter the cost of meeting interna- Tanzania. Selected presentation at the
tional standards for a nutritious diet. annual meetings of the American Society
for Nutrition (ASN), 12 June. Boston:
Tufts University.
Supplementary Material Beatty, T.K., B.H. Lin, and T.A. Smith. 2014.
Is Diet Quality Improving?
Supplementary material are available at Distributional Changes in the United
American Journal of Agricultural Economics States, 1989–2008. American Journal of
online. Agricultural Economics 96 (3): 769–89.
US dollars per day (at 2011 PPP prices)
Downloaded from https://academic.oup.com/ajae/article-abstract/100/5/1285/5073250 by University of Ghana. Balme Library user on 24 July 2019
1300 October 2018 Amer. J. Agr. Econ.
Bekkers, E., M. Brockmeier, J. Francois, and Programming: Impact of Achievement
F. Yang. 2017. Local Food Prices and Functions. European Journal of Clinical
International Price Transmission. World Nutrition 69 (11): 1272–8.
Development 96: 216–30. Green, R., L. Cornelsen, A.D. Dangour, R.
Chastre, C., A. Duffield, H. Kindness, S. Turner, B. Shankar, M. Mazzocchi, and
LeJeune, and A. Taylor. 2007. The R.D. Smith. 2013. The Effect of Rising
Minimum Cost of a Healthy Diet. Food Prices on Food Consumption:
London: Save the Children UK. Systematic Review with Meta-
Clements, K.W., and J. Si. 2018. Engel’s Regression. BMJ (Clinical Research Ed.)
Law, Diet Diversity and the Quality 346: f3703.
of Food Consumption. American Håkansson, A. 2015. Has it Become
Journal of Agricultural Economics 100 Increasingly Expensive to Follow a
(1): 1–22. Nutritious Diet? Insights from a New
Cofer, E., E. Grossman, and F. Clark. 1962. Price Index for Nutritious Diets in
Family Food Plans and Food Costs: For Sweden 1980-2012. Food & Nutrition
Nutritionists and Other Leaders Who Research 59 (1): 26932.
Develop or Use Food Plans. Home Harttgen, K., S. Klasen, and R. Rischke.
Economics Research Report No. 20. 2016. Analyzing Nutritional Impacts of
Washington, DC: USDA, Agricultural Price and Income Related Shocks in
Research Service. Malawi: Simulating Household
Cost of Nutritious Diets Consortium. 2018. Entitlements to Food. Food Policy 60:
Indicators and Tools for the Cost of 31–43.
Nutritious Diets. Boston, MA: Tufts Herforth, A. 2017. Indicators of Affordability
University, 13. of Nutritious Diets. Presentation and
Deptford, A., T. Allieri, R. Childs, C. Damu, learning laboratory at the Agriculture
E. Ferguson, J. Hilton, P. Parham, et al. Nutrition and Health Academy week,
2017. Cost of the Diet: A Method and Kathmandu, July 2017.
Software to Calculate the Lowest Cost of IFPRI. 2017. HarvestChoice. Washington,
Meeting Recommended Intakes of DC: IFPRI. Available at: www.ifpri.org/
Energy and Nutrients from Local Foods. project/harvestchoice (last accessed July
BMC Nutrition 3 (1): 26. 17, 2018).
Diewert, W.E. 1993. The Early History of Institute of Medicine. 2006. Dietary
Price Index Research. Chapter 2 in Reference Intakes: The Essential Guide to
Essays in Index Number Theory, Vol. I., Nutrient Requirements. Washington, DC:
ed. W.E. Diewert and A.O. Nakamura The National Academies Press.
33–65. Amsterdam: Elsevier. Jensen, R.T., and N.H. Miller. 2010. A
Diewert, W.E., J. Greenlees, and C.R. RevealedPreference Approach to
Hulten (eds.) 2010. Price Index Concepts Measuring Hunger and Undernutrition.
and Measurement. Chicago, IL: NBER NBER Working Paper 16555.
and University of Chicago Press. Cambridge, MA: National Bureau of
FAO. 1996. Rome Declaration on World Economic Research.
Food Security and World Food Summit Maillot, M., F. Vieux, F. Delaere, A. Lluch,
Plan of Action. Rome: FAO. and N. Darmon. 2017. Dietary Changes
———. 2018. Food Price Index. Rome: FAO. Needed to Reach Nutritional Adequacy
Available at: www.fao.org/worldfoodsitu- without Increasing Diet Cost According
ation/foodpricesindex (last accessed July to Income: An Analysis among French
17, 2018). Adults. PLoS One 12 (3): e0174679.
FAO and FHI360. 2016. Minimum Dietary Martin-Prevel, Y., M. Arimond, P. Allemand,
Diversity for Women-a Guide to D. Wiesmann, T. Ballard, M. Deitchler,
Measurement. Rome: FAO. M.C. Dop, et al.; on behalf of the
Feed the Future. 2018. Feed the Future Women’s Dietary Diversity Project
Indicator Handbook. Available at: (WDDP) Study Group. 2017.
https://www.agrilinks.org/post/feed-future- Development of a Dichotomous Indicator
indicator-handbook (last accessed July 17, for Population-Level Assessment of
2018). Dietary Diversity in Women of
Gerdessen, J.C., and J.H.M. De Vries. 2015. Reproductive Age. Current Developments
Diet Models with Linear Goal in Nutrition 1 (11): e001701.
Downloaded from https://academic.oup.com/ajae/article-abstract/100/5/1285/5073250 by University of Ghana. Balme Library user on 24 July 2019
Masters et al. Measuring the Affordability of Nutritious Diets in Africa 1301
MoFA. 2017. Agribusiness Unit Production Stadlmayr, B. 2012. West African Food
Guide: Mango Production. Accra: Composition Table. Rome: FAO.
Ministry of Food and Agriculture. Stigler, G.J. 1945. The Cost of Subsistence.
Available at: https://mofa.gov.gh/site/? American Journal of Agricultural
page_id¼14124 (last accessed July 17, Economics 27 (2): 303–14.
2018). UNICEF. 2016. From the First Hour of Life.
Nortey, John. 2017. Monitoring the Cost of New York: UNICEF.
Nutritious Diets: Ghana as a Pioneer. USDA. 2013. National Nutrient Database for
Presentation at the Committee on World Standard Reference Release 28.
Food Security Event on Impact Available at: https://ndb.nal.usda.gov/
Assessment of Policies to Support ndb/search/list (last accessed July 17,
Healthy Food Environments and 2018).
Healthy Diets, May 2017. Rome: FAO. ———. 2017. Food Plans: Cost of Food.
O’Brien-Place, P.M., and W.G. Tomek. 1983. Washington, DC: Center for Nutrition
Inflation in Food Prices as Measured by Policy and Promotion. Avaialble at:
Least-Cost Diets. American Journal of https://www.cnpp.usda.gov/USDAFood
Agricultural Economics 65 (4): 781–4. PlansCostofFood (last accessed July 17,
Omiat, G., and G. Shively. 2017. Charting the 2018).
Cost of Nutritionally-Adequate Diets in Vossenaar, M., F.A. Knight, A. Tumilowicz,
Uganda, 2000-2011. African Journal of C. Hotz, P. Chege, and E.L. Ferguson.
2017. Context-Specific Complementary
Food, Agriculture, Nutrition and
Feeding Recommendations Developed
Development 17 (01): 11571–91.
Using Optifood Could Improve the Diets
Optifood. 2012. Optifood: An Approach to
of Breast-Fed Infants and Young
Improve Nutrition. NCT01646710.
Children from Diverse Livelihood
ClinicalTrials.gov. Available at: https://
Groups in Northern Kenya. Public
clinicaltrials.gov/ct2/show/NCT01646710
Health Nutrition 20 (06): 971–83.
(last accessed July 17, 2018). World Bank. 2017a. International
Parlesak, A., I. Tetens, J.D. Jensen, S. Smed, Comparison Program (ICP).
M.G. Blenkus, M. Rayner, N. Darmon, Washington, DC: World Bank. Available
and A. Robertson. 2016. Use of Linear at: www.worldbank.org/en/programs/icp
Programming to Develop Cost- (last accessed July 17, 2018).
Minimized Nutritionally Adequate ———. 2017b. World Development
Health Promoting Food Baskets. PLoS Indicators. Washington, DC: World
One 11 (10): e0163411. Bank. Available at: http://data.world-
Rippy, D. 2014. The First Hundred Years of bank.org/products/wdi (last accessed July
the Consumer Price Index: A 17, 2018).
Methodological and Political History. World Food Programme. 2017. Findings for
Washington, DC: U.S. Bureau of Labor IANDA/Cost of the Diet in Ghana.
Statistics. Rome: WFP. Available at: https://docs.
Shiraseb, F., F. Siassi, M. Qorbani, G. wfp.org/api/documents/743ba945c9b54b6
Sotoudeh, R. Rostami, E. Narmaki, P. 68643d77786b71b0a/download (last
Yavari, M. Aghasi, and O.M. Shaibu. accessed July 17, 2018).
2016. Higher Dietary Diversity is World Health Organization (WHO) and
Related to Better Visual and Auditory UNICEF. 2007. Indicators for Assessing
Sustained Attention. British Journal of Infant and Young Child Feeding
Nutrition 115 (08): 1470–80. Practices. Geneva: WHO and UNICEF.
Downloaded from https://academic.oup.com/ajae/article-abstract/100/5/1285/5073250 by University of Ghana. Balme Library user on 24 July 2019