The latest report on The State of Food Security and Nutrition in the World confirms that the challenges of ending hunger, food insecurity and all forms of malnutrition – especially stunting, undernutrition, micronutrient deficiencies, overweight and obesity – continue to emerge(1). Although malnutrition in the population is the result of various factors, it is often fundamentally related to inadequate intake of essential nutrients due to limited income(Reference Hirvonen, Bai and Headey2,Reference Bai, Alemu and Block3) . Recently, several factors have contributed to exacerbating the lack of access to sufficient and nutritious food: global health emergencies such as COVID-19(Reference Lasdun, Harou and Magomba4); geopolitical events like the Russia-Ukraine conflict(Reference Lin, Li and Jia5) and climate change, along with climate-related shocks. Additionally, high inflation rates, particularly affecting food prices(Reference Vos, Glauber, Hernández, McDermott and Swinnen6), disproportionally impact Cali’s poorer households, which already allocate a larger share of their budgets to food (Engel’s Law) and are primarily net buyers(Reference Aksoy and Isik-Dikmelik7,Reference Ivanic and Martin8) , making them especially vulnerable to fluctuations in food prices.
Recent studies propose assessing economic access to sufficient and nutritious food by measuring affordability across different levels of diet quality at a minimum cost. Commonly used metrics include (i) the cost of caloric adequacy (CoCA), which is the minimum cost of a diet that provides adequate calories to cover the individual’s daily estimated energy requirement (EER)(Reference Mahrt, Mather and Herforth9,Reference Kachwaha, Nguyen and DeFreese10) ; (ii) the cost of nutrient adequacy (CoNA), which is a least-cost nutrient-adequate diet that, in addition to satisfying the EER, provides adequate levels of macro- and micronutrients within their minimum and maximum limits to prevent deficiencies and avoid toxicity(Reference Deptford, Allieri and Childs11–Reference Biehl, Klemm and Manohar15) and (iii) the cost of a recommended diet (CoRD) or a healthy diet, which is a diverse least-cost diet that adheres to the food group recommendations outlined in Colombia’s national Food-Based Dietary Guidelines (FBDG)(Reference Dizon, Herforth and Wang16,Reference Herforth, Venkat and Bai17) .
A prevalent trend in the literature is to estimate the minimum cost of various diets for specific populations, typically a median healthy woman of reproductive age(Reference Hirvonen, Bai and Headey2,Reference Masters, Bai and Herforth12,Reference Herforth, Bai and Venkat18,19) . To the best of our knowledge, except for a recent global study on CoNA estimates(Reference Bai, Herforth and Masters20), there has been limited research on how demographic factors influence these cost estimates. Similarly, affordability analyses of least-cost diets predominantly use representative units and individualised diets, which specify dietary composition for each individual or representative household member(Reference Herforth, Bai and Venkat18,19,Reference Geniez, Mathiassen and De Pee21,Reference Giacobone, Tiscornia and Guarnieri22) . More recently, a new approach has emerged for assessing the affordability of shared diets. These diets are optimised to be shared among household members, with individual quantities adjusted to each member’s energy requirements while considering the combined nutrient requirements of all household members, accounting for the nutrient requirements of the neediest member(Reference Schneider, Christiaensen and Webb23).
Most studies examining one or two of the three least-cost diet metrics have been conducted primarily in South Asia(Reference Kachwaha, Nguyen and DeFreese10,Reference Biehl, Klemm and Manohar15,Reference Dizon, Herforth and Wang16,Reference Geniez, Mathiassen and De Pee21) , Southeast Asia(Reference Mahrt, Mather and Herforth9,Reference Baldi, Martini and Catharina24) and East Africa(Reference Schneider, Christiaensen and Webb23,Reference Frega, Lanfranco and De Greve25) , with fewer focusing on Latin America(Reference Biehl, Klemm and Manohar15,Reference Giacobone, Tiscornia and Guarnieri22) . Their findings are not easily applicable to Latin American contexts, such as Colombia’s, or to local city contexts with high levels of economic inequality, such as Cali (Gini: 0·512), whose food system is characterised by a lack of agricultural vocation, except for small-scale production and sugarcane crops. Furthermore, the city’s supply system has demonstrated vulnerability to exogenous events such as climatic and environmental shocks and social disturbances(Reference Rankin, Hurtado and Bonilla-Findji26), highlighting its dependence on stable market conditions. This is evidenced by the 2021 Colombian protests during the COVID-19 pandemic, which were triggered by a failed tax-reform proposal and led to supply-chain disruptions and shortages of basic goods, including food.
In this context, using data for September 2022 gathered from Cali’s urban population, this study aims to determine the minimum cost and affordability of caloric-adequate, nutrient-adequate and healthy diets for urban households in the city. First, employing a differentiated approach, the study estimates the minimum cost of these least-cost diets according to age, sex and physiological condition. Second, our study performs an affordability analysis by adopting a representative household approach with individualised diets. This analysis includes calculating two measures: the ratio of per capita diet costs to average household per capita food expenditure and the proportion of households unable to afford either of the least-cost diets.
Data and methods
Data
Retail prices of locally available foods and their nutrient composition
Data on wholesale prices and supply were obtained from the Information System for Prices and Supply of the Agricultural Sector (SIPSA)(27), which captures information from the Valle del Cauca Supply Centre (CAVASA) and four wholesale satellite marketplaces in Cali. The locally available foods were those with a monthly supply in kilograms above the first quartile of the distribution, reflecting seasonal supply patterns. Food items with very low caloric intake, such as condiments, herbs, spices and extracts, were excluded.
Based on data from a large local supermarket for the third quarter of 2022, retail prices of locally available foods were calculated by applying the retailer’s mark-up to wholesale prices, with mark-ups assumed to be constant for foods within the same category. Finally, retail prices were expressed per 100 grams of the edible portion (see online supplementary material, Supplemental Table S1).
Energy and nutrient content of the foods consumed by the country’s population was assigned according to the Colombian Food Composition Table(28). The nutrient composition included macronutrients (proteins, lipids and carbohydrates) and thirteen micronutrients (Ca, Zn, Fe, Mg, phosphorus, vitamin C, thiamine, riboflavin, niacin, folate, vitamin B12, vitamin A and Na). Information gaps for certain nutrients were addressed through food homologation, imputation by similarity using the most reliable available sources or by borrowing data from the sole available source(29).
Estimated energy and essential nutrient requirements of Cali’s urban population
The EER was calculated based on information about population-specific characteristics, such as average weight, height and predominant physical activity levels for each age group and sex in the region. For individuals under 65 years of age, these characteristics were obtained from the National Survey of Nutritional Situation in Colombia (ENSIN 2015)(30). For those aged 65 years and older, information was sourced from Colombia’s Health, Well-being and Ageing Study (SABE)(31). EER for children was estimated under Resolution 3803 of 2016 of the Colombian Ministry of Health and Social Protection, which established the Recommendations for Energy and Nutrient Intake (RIEN)(32). To account for population-specific characteristics, equations described in the dietary reference intakes by the Institute of Medicine(33) were used to estimate adults’ EER. For pregnant and lactating women, additional caloric needs during the third trimester of pregnancy and the first six months (postpartum) of exclusive breastfeeding were added to the baseline requirements for non-pregnant and non-lactating women of the same age group. Finally, to aggregate across age groups, a weighted average of the EER was computed based on the projections of the National Population and Housing Census, with weights reflecting the population distribution by sex and age.
To establish the minimum and maximum macronutrient requirements by age group, the acceptable macronutrient distribution range was used. For micronutrients, the lower limit corresponded to the Estimated Average Requirement (EAR) that meets the requirements of 50 % of healthy individuals, while the tolerable upper intake level represented the highest intake level needed to avoid the risk of adverse health effects for the majority of the population. For Na, Adequate Intake (AI) was considered as the minimum reference value and the upper intake level as the maximum. For nutrients without an upper intake level – namely, riboflavin, thiamine and vitamin B12 – only the estimated average requirement was used.
Estimation of the minimum costs of diets
Cost of a caloric-adequate diet
The CoCA is estimated by selecting the food – or set of foods – that provides the number of calories necessary to satisfy the EER. Based on prior research(Reference Bai, Alemu and Block3,Reference Mahrt, Mather and Herforth9) , in a given demographic group
$i \in {\cal N}$
, the CoCA has been calculated using the following linear programming model:
$CoC{A^{\left( i \right)}} = \mathop {Min}\limits_{x_1^{\left( i \right)} \ldots \;x_J^{\left( i \right)}} \;\mathop \sum \limits_{j = 1}^J {p_j}x_j^{\left( i \right)}$
subject to
$\mathop \sum \limits_{j = 1}^J {e_j}x_j^{\left( i \right)} = EE{R^{\left( i \right)}}$
${\forall _{j \in \left\{ {1, \ldots, J} \right\}}}:\;x_j^{\left( i \right)}\; 0$
where
$\;x_j^{\left( i \right)} \in {\mathbb{R}^{ + + }}$
is the quantity – expressed in grams – of food
$j \in \left\{ {1,\; \ldots, \;J} \right\}$
for an individual in the i-th demographic group;
$\;{p_j} \in {\mathbb{R}^{ + + }}$
and
$\;{e_j} \in {\mathbb{R}^{ + + }}$
correspond to the retail price and energy content of the j-th food, respectively. The EER for an individual in demographic group
$i \in {\cal N}$
is expressed as
$EE{R^{\left( i \right)}} \in {\mathbb{R}^{ + + }}$
.
The optimal solution to the linear programming model is to construct a diet composed of food whose price-per-kilocalorie is minimal. Thus, the optimal solution follows the form
$\left( {\tilde x_1^{\left( i \right)},\; \ldots, \tilde x_m^{\left( i \right)}, \ldots, \tilde x_J^{\left( i \right)}} \right) = \left( {0,\; \ldots, EE{R^{\left( i \right)}}/{e_m}, \ldots, 0} \right)$
, where for some
$m \in \left\{ {1,\; \ldots, \;J} \right\}$
,
${p_m}/{e_m} = \;mi{n_j}\left\{ {{p_j}/{e_j}} \right\}$
. This means selecting the quantity of the least-cost starchy staple food needed to meet the EER.
Cost of nutrient-adequate diet
The CoNA is determined by the set of available foods that, for the location and period of study, satisfy the EER at the lowest cost, as well as the lower and upper limits of macro- and micronutrient intake. For a demographic group
$i \in {\cal N}$
, following extensions of Stigler’s seminal article(Reference Paris34), the CoNA metric is obtained as the solution to the following linear programming model
$CoN{A^{\left( i \right)}} = \mathop {\\min }\limits_{x_1^{\left( i \right)} \ldots \;x_J^{\left( i \right)}} \mathop \sum \limits_{j = 1}^J {p_j}x_j^{\left( i \right)}\;$
subject to
$\mathop \sum \limits_{j = 1}^J {e_j}x_j^{\left( i \right)} = EE{R^{\left( i \right)}}$
${\forall _{k \in \left\{ {1, \ldots, K} \right\}}}:\;\mathop \sum \limits_{j = 1}^J {a_{jk}}x_j^{\left( i \right)}\; \le\;U_k^{\left( i \right)}$
${\forall _{k \in \left\{ {1, \ldots, K} \right\}}}:\;\mathop \sum \limits_{j = 1}^J {a_{jk}}x_j^{\left( i \right)}\; \;L_k^{\left( i \right)}$
${\forall _{j \in \left\{ {1, \ldots, J} \right\}}}:\;x_j^{\left( i \right)}\; 0$
Given a demographic group,
$\;i \in \left\{ {1,\; \ldots, n} \right\}$
, and a nutrient,
$\;k \in \left\{ {1, \ldots, K} \right\}$
,
$L_k^{\left( i \right)} \in {\mathbb{R}^{ + + }}$
and
$\;U_k^{\left( i \right)} \in {\mathbb{R}^{ + + }}$
express the lower and upper limits of the k-th nutrient for the i-th demographic group, respectively; for all
$\;j \in \left\{ {1,\; \ldots, J} \right\}$
,
${a_{kj}} \in {\mathbb{R}^ + }$
corresponds to the content of the k-th nutrient in the j-th locally available food.
This linear programming model facilitates the identification of limiting nutrients, which are those potentially at low levels(Reference Frega, Lanfranco and De Greve25,Reference Briend, Darmon and Ferguson35) . Formally, a nutrient
$\;p$
is considered limiting if, given the optimal solution
$\left( {\tilde x_1^{\left( i \right)},\; \ldots, ,\tilde x_J^{\left( i \right)}} \right)$
, its intake level exactly meets the lower limit, i.e.
$\;\mathop \sum \nolimits_{j \in \left\{ {1, \ldots, \;J} \right\}} {a_{jp}}\tilde x_j^{\left( i \right)} = L_p^{\left( i \right)}$
.
Cost of recommended diet
The CoRD is determined by the lowest-cost set of foods that meet the recommendations for food groups specified in the FBDG for the Colombian population(36). These guidelines provide the required number of food exchanges for various demographic groups, adjusted for both EER by age group and caloric intake by food group (see online supplementary material, Supplemental Table S2).
Calculating the CoRD is simpler than estimating the CoNA. It involves straightforward methods such as selecting the least-cost food items from various food groups. The number selected from each group is determined by two criteria: (i) at least one item per food group to ensure intergroup diversity and (ii) 2–4 items per group, when the recommendation exceeds two exchanges, to ensure intragroup diversity(Reference Dizon, Herforth and Wang16,Reference Herforth, Bai and Venkat18) . Consequently, our study selected 12 least-cost items, as detailed in Table 1. These items were chosen based on their price per edible serving, calculated by dividing the price per 100 grams of the edible portion by 100 and multiplying by the serving size (in grams) specified in the dietary guidelines.
Foods selected by food group or subgroup based on Colombia’s food-based dietary guidelines (GABA)
GABA = Food-Based Dietary Guidelines. Source: Own calculations based on information from GABA. *Kumis = traditional fermented cow’s milk drink.
After selecting the twelve least-cost food items within each food group, the quantity of each item in the healthy diet adheres to two criteria: (i) all items within a group are included in the same quantity and (ii) each group must provide the number of food exchanges specified by the dietary recommendations for each demographic group, as outlined in online supplementary material, Supplemental Table S2. It is straightforward to show that these criteria align with established methods for estimating the CoRD, as described in prior research(Reference Dizon, Herforth and Wang16,Reference Herforth, Venkat and Bai17) , which defines the CoRD as the average price per edible serving for each food group multiplied by the number of servings recommended for that group.
Affordability indicators
The affordability of least-cost diets for 885 urban households, based on the Great Integrated Household Survey and using expansion factors, was assessed through a representative household approach with individualised diets. This means the total minimum cost of a diet for each household was calculated as the sum of the estimated minimum cost for each member. According to the Great Integrated Household Survey published in September 2022, the average urban household size in Cali has approximately three members, with 72 % of these being nuclear households(37). By verifying household composition by age, the representative household is defined as follows: a moderately active woman and man aged 31–50 years and a moderately active girl aged 9–13 years.
For urban households in Cali, the distribution of monthly per capita income was derived from the September 2022 Great Integrated Household Survey, and the share of household food expenditure was calculated using data from the 2022 Quality-of-Life Survey, conducted annually. Following Engel’s law, this share is calculated across various income levels, specifically for each percentile of the per capita household income distribution (Table 2). The positively skewed distribution indicates a significant concentration of households at lower income levels, suggesting potential economic limitations in access to food for a substantial proportion of urban households.
Current income and monthly food expenditure per capita by household by percentiles of per capita income
Source: Own calculations based on information from the Great Integrated Household Survey (GIHS) and Quality-of-Life Survey (QLS).
Two measures of affordability are computed at each income level: first, the ratio of the estimated per capita minimum cost of a diet for the representative household to the average per capita household expenditure on food; and second, the proportion of households unable to afford each least-cost diet. If the cost-to-expense ratio is
$ \gt 1\;$
, it reveals how many times the diet is more expensive than the average per capita food expenditure, while a ratio of
$ \le 1\;$
indicates the proportion of the average per capita food expenditure required to access the least-cost diet.
Results
Estimated minimum cost of diet
Estimated daily cost of caloric adequacy
The linear programming model used to estimate the CoCA metric indicates that the optimal solution is the quantity of the cheapest starchy staple food – in this case, a variety of rice priced at 0·31 USD (1192·07 COP) per 1000 kcal – that meets the EER. This price and the individuals’ EER together directly determine the daily CoCA results. Our findings reveal that the trajectory of the estimated CoCA was similar for women and men aged 1–18 years. For women, the maximum CoCA value was reached in the 14–18 age group at 0·87 USD (3278·2 COP), and for men, the maximum CoCA value was reached in the 19–30 age group at 0·89 USD (3379·8 COP). The minimum CoCA value was found in the 1–3 age group for both sexes. The estimated CoCA of pregnant women and lactating mothers was lower compared with that of men and was significantly higher for men than for women in the 18–50 age group (Figure 1 and see online supplementary material, Supplemental Table S3).
Cost of caloric adequacy per day. The daily CoCA (USD) estimates were differentiated by age, sex, and physiological condition. The dashed line represents the weighted average daily CoCA (0·71 USD (2680·5 COP)). Source: Own calculations based on the information of the study.
Estimated cost of nutrient adequacy
Estimated daily cost of nutrient adequacy and cost of nutrient adequacy per 1000 kcal
The trajectory of the daily CoNA was similar for women and men aged 4–18 years. In the 14–18 age group, the estimated daily CoNA reached its maximum value for men at 1·91 USD (7228·2 COP) and women at 1·87 USD (7088·3 COP). Among pregnant women and lactating mothers, the lowest daily CoNA was found in lactating mothers aged 31–50 years (1·64 USD (6223·2 COP)), and the highest daily CoNA was found in pregnant women under 18 years of age (2·07 USD (7845·5 COP)) (Figure 2(a); for values, see online supplementary material, Supplemental Table S4).
Cost per day and cost per 1000 kcal for a nutrient-adequate diet. The CoNA estimates were differentiated by age, sex, and physiological condition. (a) Estimated daily CoNA (USD). (b) Estimated CoNA per 1000 kcal. (USD) The dashed line represents (a) the weighted average daily CoNA (1·66 USD (6266·16 COP)) and (b) the average CoNA per 1000 kcal (0·76 USD (2861·1 COP)). Source: Own calculations based on the information of the study.
In contrast to the daily CoNA, the CoNA per 1000 kcal was calculated to adjust for differences across energy requirements. The median CoNA per 1000 kcal was calculated to be 0·75 USD (interquartile range: 0·68–0·83). The highest levels were observed in girls aged 4–8 years and women aged 51–69 years (Figure 2(b); for values, see online supplementary material, Supplemental Table S4). Notably, among individuals aged 18 years and older, women had a higher CoNA per 1000 kcal than men, indicating a greater need for a nutrient-dense diet. In addition, younger individuals aged 4–13 years and adults aged 31 years and older exhibited a markedly higher CoNA per 1000 kcal compared with other age groups.
Limiting nutrients
Estimating the CoNA through a linear programming model makes it possible to identify limiting nutrients – i.e. those that significantly constrain the estimated minimum cost of the diet. This study reveals that, across demographic groups, vitamin C, vitamin A, vitamin B12 and Ca are the vitamins and minerals that significantly constrain the estimated CoNA (Table 3). An analysis of specific demographic groups reveals that CoNA is highly sensitive to Fe requirements in pregnant women. Zn is particularly costly for men, except those aged 9–13 years, and for girls aged 8 years or younger, as well as those aged 51 years or older. Additionally, Zn requirements also restrict estimated CoNA for lactating women over 18 years. Similarly, Mg influences CoNA for individuals aged 8 years or younger and those aged 31 years or older, regardless of sex.
Limiting nutrients in the least-cost nutritious diet
For any nutrient, the difference (%) between the optimal CoNA contribution and the minimum level of intake needed for the nutrient considered is reported. Source: Own calculations.
Estimated daily cost of recommended diet
Figure 3 shows that for women and girls, the highest CoRD was found in adolescents aged 14–18 years (2·83 USD (10 697·40 COP)), while for men and boys, it was found in adults between 19–30 years (2·98 USD (11 284·4 COP)). Among pregnant women and lactating mothers, the lowest CoRD was registered among pregnant women under 18 years of age (2·5 USD (9455·5 COP)), while the highest CoRD was registered among lactating mothers aged 19–30 years (2·71 USD (10 254·8 COP)). Greater variations in the CoRD can be seen across age groups, and the difference in the minimum costs between men and women in the 19–30 age group (2·98 USD (11 284·38 COP) v. 2·10 USD (7931·35 COP)) and the 31–50 age group (2·82 USD (10 663·4 COP) v. 1·98 USD (7477·8 COP)) are noticeable. (For details, see online supplementary material, Supplemental Table S5.)
Cost per day of a recommended diet. The CoRD estimates were differentiated by age, sex, and physiological condition. The dashed line represents the weighted average CoRD (2·34 USD (8860·11 COP)). Source: Own calculations based on the information of the study.
Affordability of a caloric-adequate diet, a nutrient-adequate diet and a healthy diet
Indicator 1: Affordability for urban households
Using the approach of individualised diets for a representative household, Table 4 presents the minimum daily costs for three levels of increasing diet quality. These costs are differentiated by the age group and sex of each household member and include the CoRD:CoCA and CoRD:CoNA ratios. The minimum costs associated with each type of diet and the information on the monthly per capita food expenditure per urban household make it possible to identify significant economic barriers to achieving certain dimensions of food security(Reference Herforth, Venkat and Bai17). Approximately 42·66 % of households in the 10th percentile cannot afford the CoCA, accounting for 4·27 % of all urban households in the city. Additionally, in the 10th–20th percentiles, 100 % of households cannot afford the CoNA. In the 30th percentile, 40·97 % cannot afford the CoNA. Thus, approximately 24·10 % of households cannot afford a diet that provides the necessary levels of calories and essential nutrients. Lastly, all households in the 10th–30th percentiles cannot afford the CoRD, while approximately 39·23 % of households in the 40th percentile can afford this type of diet. Approximately 36·08 % of urban households in Cali cannot afford a diverse diet that not only meets the national dietary recommendations but also promotes long-term health (Figure 4).
Least-cost diet by three increasing levels of diet quality for representative households
Source: Own calculations based on the information of the study.
Proportion (%) of urban households in Cali, Colombia, that cannot afford any of the three diet types. The proportion of households, differentiated by percentiles of per capita income, whose per capita expenditure on food is lower than the per capita cost of each diet type. The white area represents the proportion of households (63·92 %) that can afford the three diet types. Source: Own calculations based on the information of the study.
Indicator 2: Cost-to-expense ratio
Figure 5 shows the indicator of the ratio of the estimated per capita minimum cost of a diet for households to the average per capita expenditure on food by households at each income level. The per capita CoCA is 1·01 times the average per capita food expenditure for households in the 10th percentile, 26 % of spending for households in the 50th percentile and approximately 15 % for those in the 80th percentile. This observation verifies the limitations of households in the 10th percentile in terms of their ability to afford a subsistence diet. The CoNA is more than twice the average per capita food expenditure for households in the 10th percentile and 1·18 times for households in the 20th percentile. Similarly, there is a lack of affordability for households between the 10th and 40th percentiles of the CoRD. For households in the 50th percentile, the affordability of a healthy diet is likely to be constrained, as the per capita CoRD represents approximately 87 % of their average per capita food expenditure.
Ratio of the per capita cost to the average per capita expenditure on food. The ratio of the estimated per capita minimum cost (USD) of a diet for households to the average per capita expenditure on food (USD) by households at each income level. Source: Own calculations based on the information of the study.
Affordability outcomes are sensitive to the share of household food spending. To provide a conservative lower-bound estimate, we assume urban households allocate their entire income to food. Under this assumption, only households in the lowest income group (10th percentile) face challenges in affording the least-cost diets: 13·71 % cannot afford the CoCA; 42·22 % the CoNA and over 67 % a healthy diet. The cost-to-expense ratio shows that the per capita CoNA accounts for 93 % of the average per capita food expenditure for this group. Therefore, even if all income were allocated to food, the per capita CoRD would still exceed the average per capita food expenditure by 1·29 times.
Discussion
This study determines the cost and affordability of three levels of diet quality for urban households in Cali, Colombia, using least-cost diet metrics. We identified significant cost variations across demographic groups categorised by age, sex and physiological condition. Adhering to Colombia’s FBDG recommendations requires a premium to meet energy and nutrient requirements. Only 63·92 % of urban households in Cali can afford a healthy diet, revealing severe economic barriers to affordable nutrition for over 30 % of the population. The cost-to-expense ratio further demonstrates that reallocating expenditures towards diet is insufficient to close the affordability gap for lower-income households. These key findings are discussed in detail below.
First, our study reveals notable variations in least-cost diet metrics across demographic groups. Consistent with previous research(Reference Bai, Alemu and Block3,Reference Mahrt, Mather and Herforth9) , the CoCA metric is calculated by multiplying the price per kilocalorie of the least-cost starchy staple – in this case, a specific variety of rice – by the individual’s EER. As expected, the daily CoCA is notably higher for men in the 19–30 group due to their increased energy needs. Unlike studies that estimate the CoNA based on a representative unit – e.g. agent(Reference Hirvonen, Bai and Headey2,Reference Herforth, Bai and Venkat18,Reference Briend, Darmon and Ferguson35) or household(Reference de Pee, Hardinsyah and Jalal14,Reference Biehl, Klemm and Manohar15,Reference Geniez, Mathiassen and De Pee21,Reference Giacobone, Tiscornia and Guarnieri22,Reference Baldi, Martini and Catharina24) – our analysis accounts for the influence of demographic and physiological factors on dietary requirements, revealing significantly a higher daily CoNA for adolescents and lactating or pregnant women than for other consumers. These barriers to access, particularly for the latter groups, constitute a substantial public health concern, as they may increase the risk of adverse pregnancy outcomes and contribute to developmental delays and cognitive impairments in children(Reference Gebre, Biadgilign and Taddese38).
To account for variations in daily CoNA due to differences in EER across demographic groups, we also express the CoNA on a per 1000 kcal basis. Our results, which are in line with a previous recent study(Reference Bai, Herforth and Masters20), reveal an N-shaped trajectory in CoNA per 1000 kcal and that CoNA per 1000 kcal, for individuals aged 19 years and older, is generally higher for women than for men, indicating that a diet with a higher density of nutrients with locally available foods tends to be more costly for women.
Our study indicates that a healthy diet ensures nutrient adequacy with some variability, meeting about 83·2 % of the lower limits of nutrient requirements (±10 %) (see online supplementary material, Supplemental Table S6 for CoRD validation). This finding corresponds to prior research showing that such diets typically fulfill over 80 % of nutrient requirements on average(Reference Van, Herforth and Trinh39), and it is further corroborated by a global study reporting 89 % (±5 %) fulfillment of nutrient needs(Reference Herforth, Venkat and Bai17). Notably, adolescent girls aged 14–18 years and adult men of 19–30 years have to deal with a higher daily CoRD, driven by the cost of meeting dietary recommendations for two specific food groups: (i) meat, eggs, legumes, nuts and seeds and (ii) milk and dairy products. These two groups, along with the fruits and vegetables group, represent around 70 % of the total cost of a healthy diet (Figure 6), corroborating prior research that identifies these food groups as major contributors to the CoRD(Reference Herforth, Bai and Venkat18,Reference Van, Herforth and Trinh39) .
Share (%) of total cost for each food group in the recommended diet, across demographic groups by age, sex and physiological condition. Source: Own calculations based on the information of the study.
Second, our findings reveal the significant additional costs that urban households in Cali may incur in their efforts to improve diet quality. Specifically, the CoRD was on average 1·4 (sd: 0·18) times higher than the CoNA, indicating an additional expense associated with aligning dietary intake with the recommendations outlined in the Colombian FBDG (Table 5). This concurs with a recent study on South American countries that, using a representative agent approach, reported an identical CoRD:CoNA ratio of 1·4(40). Furthermore, our analysis revealed that CoNA exceeded CoCA by a factor of 2·4 (sd: 0·26), indicating a pronounced nutrient premium. This premium surpasses the CoNA:CoCA ratios documented in upper-middle income countries (2·18 (sd: 0·48))(Reference Herforth, Venkat and Bai17) and in the Latin American and Caribbean region (2·21 (sd: 0·73))(Reference Bai, Alemu and Block3). These results underscore the economic constraints and context-specific challenges that may hinder dietary quality improvements, nutritional adequacy achievement and the adoption of healthy dietary patterns among urban households in Cali.
Ratios of nutrient-adequate diet cost to caloric-adequate diet cost (CoNA/CoCA), recommended diet cost to caloric-adequate diet cost (CoRD/CoCA), and recommended diet cost to nutrient-adequate diet cost (CoRD/CoNA) across demographic groups by age, sex, and physiological condition
Source: Own calculations based on the information of the study.
Third, our analysis identifies vitamin C, vitamin A, vitamin B12 and Ca as the key nutrients limiting the daily CoNA across demographic groups. While this is consistent with studies in other countries(Reference Masters, Bai and Herforth12,Reference Chungchunlam, Garrick and Moughan41) , variations in local food availability, retail prices and seasonal factors have led other research to identify different limiting nutrients, such as riboflavin(Reference Frega, Lanfranco and De Greve25); Fe in children(Reference Briend, Darmon and Ferguson35); and biotin, molybdenum, potassium, selenium and pantothenic acid(Reference Chungchunlam, Garrick and Moughan41). Limiting nutrients often correlates with potential nutritional deficiencies, providing valuable insights for interventions to enhance access to micronutrient-rich foods. Long-term strategies should promote dietary diversity by supporting local production and consumption of varied, nutrient-rich foods. Policy measures, such as fiscal subsidies targeted at producers of foods that are under-consumed relative to recommended dietary levels for a healthy diet (i.e. priority foods), can enhance availability and encourage consumption(1). Additionally, medium- and short-term measures may include the fortification of mass-consumed staple foods, home fortification, biofortification and supplementation for vulnerable groups(42). Colombia has implemented food fortification strategies including fortifying wheat flour with thiamine, niacin, riboflavin, folate and Fe under Decree 1944 of 1966. Social programs also have included fortifying specific foods, such as vegetables mixes with added micronutrients, and milk and cookies with folic acid, Fe and Zn.
Fourth, our study underscores the limitations of conventional poverty measures based solely on energy standards, revealing that least-cost diet metrics offer a more comprehensive assessment of the challenges households face in achieving adequate nutrition and a healthy diet. National poverty lines are typically established by estimating the expenditure required for a consumption bundle deemed adequate for basic food and non-food needs(Reference Ravallion43). The food component establishes the extreme poverty line, based on a basket defined by the reference population’s consumption patterns and adjusted to meet the per capita daily caloric requirement. During the study period, the extreme poverty line in Cali was estimated at 58·36 USD (220 911·40 COP)(44). In comparison, the monthly per capita CoNA was 49·84 USD (188 649·10 COP), exceeding 80 % of this threshold, while the CoRD was 69·51 USD (261 114·50 COP), surpassing it by 1·2 times. These findings reveal that not only poor households but also near-poor households – classified as non-poor under conventional poverty lines – cannot afford at least a least-cost healthy diet, as confirmed by previous studies(Reference Herforth, Venkat and Bai17–19,Reference Mahrt, Herforth and Robinson45) . This has significant implications for nutrition and food security policy aimed at reducing hunger and malnutrition, as conventional poverty lines may misguide program targets by failing to capture the broader economic constraints affecting dietary quality.
Fifth, our analysis highlights both the challenges households face in affording the three increasing levels of diet quality and the difficulty in closing the affordability gap. Households in the first two quintiles have average per capita food expenditures that fall below both the monthly per capita CoNA and CoRD. This issue is especially severe for households in the 10th percentile, where even by allocating 100 % of their income to food, the per capita CoRD exceeds their average per capita food expenditure by 1·29 times (see online supplementary material, Supplemental Table S7). Therefore, closing the affordability gap cannot be achieved merely by reallocating food expenditures, i.e. without an increase in the current household income level.
Our methodology offers several key strengths in analysing affordability. First, unlike global studies that rely on country-level data(Reference Bai, Alemu and Block3,Reference Herforth, Bai and Venkat18) , we use local food availability, retail prices, nutritional recommendations, income distribution and household food expenditure, contributing to increased relevance and accuracy in our analysis. Second, our focus on the representative household, as opposed to a single representative agent, allows our affordability indicators to be sensitive to variations in household size, providing a deeper understanding of diet costs. Additionally, in line with prior research(Reference Biehl, Klemm and Manohar15,Reference Geniez, Mathiassen and De Pee21,Reference Giacobone, Tiscornia and Guarnieri22) , we employ individualised diets for each household member, which contrasts with the shared diet approach proposed in recent literature(Reference Schneider, Christiaensen and Webb23). This distinction means that our estimates offer a lower bound of household diet costs, whereas the shared diet method, in accordance with Rawls’ maximin principle, would produce upper-bound estimates by focusing on the nutritional needs of the neediest household member. This approach would likely lead to less favourable affordability outcomes, particularly for low- and middle-income economies.
Our approach has limitations. First, the locally available food set, defined from a supply-side perspective, would benefit from integrating demand-side data(Reference Starck, Blumfield and Keighley46,Reference Lauk, Nurk and Robertson47) on commonly consumed foods and quantities. Incorporating population consumption patterns through model constraints could better align least-cost diets with cultural dietary habits(Reference Biehl, Klemm and Manohar15,Reference Giacobone, Tiscornia and Guarnieri22) . However, the assumption that households will maintain unchanged consumption patterns while improving diet quality is problematic. For instance, evidence from Colombia indicates a significant misalignment between current dietary patterns and healthy diet recommendations, particularly among lower-income households, whose eating patterns include a higher proportion of energy-dense foods(Reference Mora-García, Ruiz-Díaz and Villegas48). Adopting a healthier diet would require shifts in consumption patterns that, though influenced by multiple factors, remain largely constrained by income. Second, assuming constant retail mark-ups within food categories for estimating retail prices, due to a lack of publicly available data in Colombia, does not fully capture retail pricing complexities, potentially introducing deviations in the affordability indicators based on the estimated least-cost diet metrics. Although useful under stable market conditions – where mark-ups are likely similar due to factors like shelf life, spoilage risks and supply stability – this assumption may overlook variations from demand elasticity and market dynamics. Previous research(Reference Darmon and Drewnowski49,Reference Timmins, Morris and Hulme50) also highlights the challenges with using national averages or prevailing retail prices for diet cost estimation. Future research should focus on dynamic pricing models to enhance accuracy. Third, the distribution of monthly per capita household income in our study, though closely aligned with Great Integrated Household Survey data used by the National Administrative Department of Statistics for poverty incidence rates, lacks supplementary administrative records. For September 2022, our estimates indicate 4·66 % of households in Cali in extreme poverty and 18·65 % in monetary poverty, compared to 5·6 % and 19·80 % reported by official statistics. Minor discrepancies may arise from the absence of additional data on pension payments and institutional aid programs, including conditional cash transfers. Similar minor discrepancies might be expected in affordability indicators.
This study highlights the significant barriers urban households in Cali, Colombia face in affording three increasing levels of diet quality. Our analysis reveals notable variations in least-cost diet metrics across demographic groups, with additional costs required to meet nutrient requirements and adhere to national FBDG recommendations. Using a representative household approach with individualised diets, this study demonstrates the substantial difficulties faced by lower-income households in affording caloric-adequate, nutrient-adequate and recommended diets, even with food expenditure reallocation. This approach provides key insights into local food system issues and informs the design and implementation of targeted public policies for the case study. Given the scarcity of similar studies in Latin America, this research establishes a valuable methodological basis for conducting analogous investigations in the region, while expanding knowledge on the food and nutritional security challenges faced by urban households in Cali.
Supplementary material
For supplementary material accompanying this paper visit https://doi.org/10.1017/S1368980025000564
Acknowledgements
We appreciate the observations and contributions made during the research by Mark Lundy, Director/Food, Environment & Consumer Behaviour, Alliance of Bioversity International and CIAT and by Margarita Gáfaro González, Manager, Banco de la República Cali Branch and Camilo Bohórquez Peñuela, Junior Researcher, Centre for Studies on Production and Sectorial Trade (CEPCO), Banco de la República Cali Branch.
We thank Olga Spellman (Alliance of Bioversity International and CIAT Science Writing Service) for technical review and copy editing of this manuscript.
Financial support
This article is derived from a research project that was part of the CGIAR Research Initiative Grant/Agreement No. G198 on National Policies and Strategies (NPS). The CGIAR launched the NPS with national and international partners to foster policy coherence, respond to crises and integrate national and subnational tools in six countries in Africa, Asia and Latin America. We thank all the funders who supported this research through their contributions to the CGIAR Trust Fund (https://www.cgiar.org/funders/).
Competing interests
The authors declare that they have no competing interests.
Authorship
A.M.Y.V., M.P.Z.L., L.E.G.C., D.E.G.G. and S.A.B.M. conceived and designed the study, contributed to the data analysis and interpretation, discussed the results and drafted and provided critical revisions to the manuscript. L.E.G.C., D.E.G.G. and S.A.B.M. conducted database collection and processing and performed the computations and analysis of the statistical data. S.R-C. and C.E.G.R. provided critical feedback on the manuscript.
Ethics of human subject participation
Not applicable.
Open access
