Article Text
PDF
Abstract
Objectives To estimate the own-price, cross-price and income elasticities for carbonated soft drinks (CSDs), malt drinks, chocolate powder, sachet water and sugar in Nigeria. These elasticities can be used to estimate the potential demand response to the recently-introduced sugar tax in Nigeria.
Setting The study uses household data from the 2018/2019 Nigeria Living Standards Survey (NLSS).
Participants The NLSS is a national household survey. 21 114 households were included in the final sample for this analysis.
Primary and secondary outcomes We used Deaton’s almost ideal demand system, which controls for the goods’ quality, to estimate the effect of price and income changes on the demand for CSDs, chocolate powder, malt drinks, sachet water and sugar.
Results We found that the own-price elasticity (ordered from most to least price-responsive) was −0.99 (p<0.01) for sachet water, −0.76 (p<0.01) for CSDs, –0.72 (p<0.01) for chocolate powder, −0.62 (p<0.01) for sugar and –0.19 (p<0.01) for malt drinks. The cross-price elasticities indicate that malt drinks and chocolate powders are substitutes of CSDs. The income elasticities indicate that all the commodities are normal goods. Sachet water had the highest income elasticity at 0.62 (p<0.01), followed by chocolate powder at 0.54 (p<0.01), CSDs at 0.47 (p<0.01), malt drinks at 0.43 (p<0.01) and sugar at 0.13 (p<0.01).
Conclusion Even though the price elasticities for CSDs, malt drinks and chocolate powder are less than one, in absolute terms, they are significantly different from zero. Increases in the sugar-sweetened beverage tax could curb the demand for these beverages, and, in turn, reduce the incidence and prevalence of sugar-attributable diseases.
- Obesity
- HEALTH ECONOMICS
- PUBLIC HEALTH
Data availability statement
Data are available in a public, open access repository. Data are available in a public, open-access repository. Data are available on request. The publicly available data can be accessed from: https://microdata.worldbank.org/index.php/catalog/3827/get-microdata.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
STRENGTHS AND LIMITATIONS OF THIS STUDY
This study bridges the evidence gap for the impact of sugar-sweetened beverage (SSB) taxes on SSB demand in low-income and middle-income countries (LMICs), using Deaton’s almost ideal demand system, with a specific focus on Nigeria.
Since we use the Living Standards Survey data, which is available for many LMICs, our study could provide a template for evaluating the impact of price and income changes on SSB demand in these LMICs.
A limitation is that we monitored SSBs consumed in the 7 days preceding data collection and excluded any households that did not consume these products in the 7 days preceding data collection.
We were not able to include all SSBs consumed in Nigeria, and only analysed the consumption of carbonated soft drinks, chocolate drinks and malt drinks.
The results are subject to self-reporting and recall biases.
Terminology
A sugar-sweetened beverage (SSB) is a non-alcoholic beverage containing sugar. Examples include carbonated and non-carbonated soft drinks (CSDs), energy drinks, sports drinks, sweetened fruit juices, sweetened drinking yoghurt, sweetened milk products, as well as others.
A CSD is a non-alcoholic beverage containing sugar and carbon dioxide. Examples of CSDs are Coca-Cola, Fanta, Sprite and others.
A soft drink is a non-alcoholic drink, that is, usually, but not necessarily, carbonated.
Malt drinks are CSDs made from malted barley, such as Maltina.
Chocolate drinks refer to chocolate powders used to make sweetened chocolate beverages, such as Milo.
Sachet water is purified water sold in plastic bags.
Introduction
The consumption of highly processed diets is driving overweightness and obesity in many countries, including Nigeria.1 2 Nigeria is a lower-middle-income country with an estimated population of 218 million. The Nigerian diet is transitioning from traditional high fibre foods to high-energy, high-fat and high-sugar processed foods.3 Nigeria has witnessed a rise in the number of fast food restaurants, accompanied by an increase in the availability of SSBs.4 Packaged juices are growing in popularity and rapidly replacing natural fruits in diets.4 These modern diets have resulted in a growing incidence of overweightness (body mass index (BMI) between 25 and 30 kg/m2), and obesity (BMI>30 kg/m2), as well as diet-related non-communicable diseases (NCDs).3 The increasing prevalence of obesity, overweightness and NCDs is a significant problem,5–8 as NCDs account for 24% of all deaths in Nigeria.9 10 The prevalence of obesity rose by 47% in men and 39% in women between 2002 and 2010.11 Between 2003 and 2018, the combined overweight and obese Nigerian population increased by 29%.12
The increased prevalence of overweightness and obesity is a global problem. It is estimated that obesity has tripled since 1975.13 In 2016, 39% of all adults aged 18 and above were overweight and 13% were obese.13
Excessive sugar consumption is linked to obesity and NCDs.14 One way to reduce the widespread prevalence of obesity and NCDs attributable to excessive sugar consumption is by making these products less affordable. An increasingly popular option is an SSB tax.15 16 SSBs include all non-alcoholic beverages that contain sugars, such as carbonated and non-CSDs, energy or sports drinks, juices and dairy. An SSB tax increases the price of SSBs, and by making them less affordable reduces their demand. This, in turn, may improve public health, and generate government revenue.17–20
Africa is an attractive market for multinational beverage companies owing to its rapid economic growth. SSB consumption increased by 72% in Africa between 2008 and 2022.21 In Nigeria, SSB consumption increased by 123% during the same period.21 Several carbonated beverage companies operate in Nigeria, but Coca-Cola (55.8%) and Pepsi (17.8%) are the leading companies in terms of market shares.21 Nigeria’s continued lack of potable water is driving the demand for bottled water, sachet water and SSBs.21 Nigeria is the second largest bottled and sachet water consumer globally.21 Among Nigerians, CSDs are the leading SSBs of choice, followed by fruit juices.21 Coca-Cola is the most popular CSD, accounting for 31% of the total CSD consumption in 2022, followed by Fanta (12%) and Sprite (6.9%).21
Nigeria introduced a tax of 10 naira (US$0.02) per litre of non-alcoholic, carbonated and sweetened beverages22 on 1 June 2022 in an effort to tackle the rising levels of obesity and NCDs, and to raise excise tax revenue which could be used for health-related and other critical expenditures.23 24
The effectiveness of taxes in curbing consumption depends crucially on the price elasticity of demand for the product. Price elasticity is the percentage change in the quantity demanded attributed to a 1% change in price.25 The demand for SSBs is also influenced by household income, the effect of which is quantified by the income elasticity of demand. The income elasticity of demand is the percentage change in demand attributed to a 1% change in income. Household expenditure is frequently used as a proxy for income as household income is often misreported in surveys.26–28 Price and income elasticities are key parameters in understanding consumers’ responses to price and income changes, and are fundamental to assessing the potential effects of the SSB tax on the consumption of these products and on government revenue.
To the best of our knowledge, there is only one published study that estimates SSB price or income elasticities in Africa (South Africa). The objective of this article is to bridge this gap and to add to the growing evidence of the price and income elasticities for SSBs in low-income and middle-income countries (LMICs). In this paper, we estimate the own-price, cross-price and income elasticities of demand for CSDs, chocolate drinks, malt drinks, sachet water and sugar, using Deaton’s almost ideal demand system (AIDS), which controls for quality differences.28 We used unit values, the quotient of expenditure and quantity, as a proxy for prices. Such estimates can be useful to other countries that are considering implementing SSB taxes.26
Data and methods
We use the Nigeria Living Standards Survey (NLSS) 2018/2019 data, which provides nationally representative living standards estimates for the 36 states and the Federal Capital Territory.29 The sampling frame for the NLSS 2018/2019 was based on the Nigeria Integrated Survey of Households 2 (NISH2). The NIHS2 contains 200 enumeration areas (EAs) per state, composed of 20 replicates of 10 sample EAs for each state, systematically drawn from all the local government areas.29 Sixty of the 200 EAs were selected for the NLSS 2018/2019 using random systematic sampling.29 Ten households were sampled per EA. This results in a sample of 600 households per state and 22 200 households overall,29 of which 22 101 households had the data required for this study.
The NLSS 2018/2019 provides information on demographics, education, labour, food and non-food expenditure, income-generating activities and other sources of household income, among other variables.
Question 2 of the food consumption expenditure questionnaire asks, ‘In total, what was the quantity of [ITEM] your household consumed in the past 7 days?’ For unit values, the questionnaire asks, ‘The most recent time your household purchased [ITEM], how much [QUANTITY] did your household purchase?’ followed by ‘How much did your household spend … on the [ITEM] purchased most recently?’29 We used the quantities of consumption reported in question 2. The quotient of expenditures and quantities reported in the most recent purchase, were used to calculate the unit values.
Since the EAs are the smallest geographical unit in the data, we used EAs to represent clusters.
We used unit values as a proxy for prices, as done by Deaton30 (see equation (1)). We used inflation-adjusted 2018/2019 naira (1 naira=US$0.002) for unit values.31
(1)
where uvGhc is the unit value of the specified good G in cluster c in household h. X is expenditure on a specified good, while q is the quantity purchased (expressed in litres or kilograms) of the specified good.
We defined individual unit value outliers as the logged unit values that deviated more than 2.5 SD from the cluster logged mean, as done by the National Bureau of Statistics (NBS).29 A total of 987 observations (4.5% of the sample) were dropped from the sample. For missing unit values (the good consumed was a gift), we used the median unit values, at the cluster level, as done by the NBS.29 A total of 21 114 households were included in our final sample.
Deaton’s AIDS
We used Deaton’s AIDS for the estimation of the demand elasticities. We used household expenditure as a proxy for household income, and estimate expenditure elasticities instead of income elasticities. Aggregate household expenditure is defined as the inflation-adjusted aggregate consumption, in monetary terms, of all foods, goods and services consumed by the household. These include education, health, dwelling and rent expenditures. Monetary values were adjusted for inflation as the data were collected over 13 months, between September 2018 and September 2019.29 32
Households may respond to price increases by switching to a lower quality, resulting in a smaller decline in the quantity consumed, than would have resulted if they had not switched to lower quality goods.27 The AIDS model controls for these quality differences.28 The AIDS approach assumes that there is no price variation within clusters, but that there is price variation between clusters.28 30 33 This assumption requires both a geographical proximity of the households in a cluster and that households are interviewed at approximately the same time.34 Thus, variations in unit values reported by households within clusters are attributed to quality variations of the goods and to reporting errors.35 36 Unit values are error-ridden since they are derived from reported expenditures and quantities, both of which are subject to measurement error.
Deaton proposes estimating two equations:
(2)
(3)
where c and h represent the cluster and household, respectively. G represents the specified good. wGhc is the proportion of specified good expenditure (budget share) to the total household expenditure. lnx is the natural log of total household monthly expenditure. z is a vector of household characteristics, including household size, age, sex and literacy level of the household head, shares of adults, males, employed adults, month of the interview, household location (rural/urban), and access to sufficient drinking water. We control for access to sufficient drinking water as a third of Nigerians do not have access to potable water.37 feGc represents an error associated with the cluster, while u0Ghc and u1Ghc are idiosyncratic errors (for each home).36 The prices of the N goods that the household can purchase are represented by πJc and are assumed not to vary in cluster c.28
lnuvGhc is the natural log of the unit value, (equation (1)), and is a function of the same variables that affect wGhc (equation (2)), except for the cluster level error (
), as adding a fixed effect to the unit value equation causes the unit values to no longer give any useful information about the prices.34 
and 
, capture the magnitude of the errors, and the relationship between 
and 
corrects the final price elasticity estimates for measurement error.38
Since the prices of the goods are not directly observed, equations (2) and (3), cannot be estimated directly. However, the assumption that prices do not vary within clusters allows them to be estimated in three stages.26 28 30 33 36 The first stage controls for the socioeconomic characteristics of households in unit values and budget shares. The effects of households’ socioeconomic characteristics are removed from households’ budget shares and unit values, and this allows us to purge the quality effects from the budget shares and unit values.35
The second stage consists of using these adjusted budget shares (from the previous stage) and unit values, averaged by cluster, to estimate errors with measurement errors (error-in-variable models) between clusters.26 36 This is then used to adjust the estimates and correct for structural correlation between quantity and unit value.39 In the third stage, the price and quality effects are disentangled relying on a separability assumption.27 35
The coefficients 
and 
represent the elasticity of budget share with respect to total expenditure (
) and the elasticity of the unit value 
with respect to total expenditure (ie, quality elasticity).28 35 The total expenditure elasticity is given by equation (4)
(4)
Similarly, these three stages allow for the estimation of total corrected price elasticities and cross-price elasticities for quality differences,28 34 36 (equation (5))
(5)
where δGJ is the Kronecker delta (equal to 1 if G=J or to 0 otherwise) and budget shares are assessed at their means.36 wG is the average share of total household expenditure for good G.
We estimated the demand elasticities conditional on positive expenditure. Households that did not purchase any goods of interest were not included in the main analysis. We test whether the results change if we take into account nil expenditures by using the inverse Mills ratio (see online supplemental appendix). We find that the results do not differ.
Supplemental material
The demand system was estimated using Stata V.17 using the duvm command.39
Patient and public involvement
Patients and the public were not involved in the design or planning of the study.
Results
The average household consists of five people (table 1). Most households (81.9%) had a male head and 59.8% of households had adults aged 18 years and above. 72.7% of households had literate household heads. The average weekly household expenditure was 17 482 naira (US$34.96). Nearly 85% of households had access to sufficient drinking water. One-third of households were located in urban areas.
Descriptive statistics (95% CIs in parenthesis)
Sugar, followed by sachet water, was purchased by the largest percentage of households (table 2). For those households that purchased these products, the average weekly purchase was 318 g of sugar and 12.4 L of sachet water. On average, households consumed 1.4 L of CSDs, 847 mL of malt drinks and 109 g of chocolate powder in the 7 days preceding data collection (table 2).
Beverage and sugar consumption in the last 7 days preceding data collection (95% CIs in parenthesis)
Table 3 shows the own-price elasticities (in bold) and cross-price elasticities. The cross-price elasticity is read as the demand response for the food item(s) in the row(s) to price changes in the food item(s) in the column(s). All own-price elasticities are negative and statistically significant. The own-price elasticity for CSDs was found to be –0.76. This implies that, on average, the demand for CSDs falls by 7.6% in response to a 10% increase in price, all else held constant. Malt drinks were found to have the lowest own-price elasticity in absolute terms, at –0.19, implying that, on average, malt drink demand falls by 1.9% in response to a 10% increase in malt drink prices, all else held constant. For chocolate powder, the own-price elasticity was found to be –0.72. Sachet water has the highest own-price elasticity (in absolute terms). The own-price elasticity for sugar was found to be –0.62 (see table 3).
Own-and-cross price elasticity estimates with quality correction and symmetry restricted
The cross-price elasticity is positive for substitutes goods, and negative for complement goods. With the exception of malt drinks and sachet water, most of the cross-price elasticities are positive, indicating that the commodities are substitutes. Sachet water was found to be a complement for all of the commodities, while malt drinks were found to be a complement for all the commodities except CSDs.
Normal goods are goods whose demand increases as income increases. All the commodities were found to have positive expenditure (income) elasticities, implying that all are normal goods. The income elasticities are between 0 and 1, indicating that demand increases less than proportionally to income (table 4). Sachet water has the highest expenditure elasticity, followed by chocolate powder, CSDs, malt drinks and sugar.
Expenditure elasticities
Discussion and conclusion
SSBs have become more affordable over time, particularly in LMICs.40 In addition, SSBs are widely available and heavily promoted.41
The WHO16 recommends the use of sugar taxation to address the effects of excessive sugar consumption. Sugar consumption is associated with obesity and NCDs.14 These are on the rise in many countries, including Nigeria.23 The Nigerian government adopted a specific tax of 10 naira (US$0.02) per litre on non-alcoholic, carbonated and sweetened beverages22 in June 2022, in an effort to tackle the rising levels of obesity and NCDs. The SSB tax will be earmarked for NCD prevention and control.42
Several studies have estimated the price elasticities for SSBs, to gauge the impact of an SSB tax on SSB demand.25 26 36 43–49 However, only one study was done in Africa, in South Africa. In nearly all cases, authors use a large household survey, with n ranging from 9 38246 to 101 651 households.45 For price elasticity estimates the technique of choice is Deaton’s AIDS, or similar approaches based on the AIDS model. The foods most often included in the estimates of SSB elasticities are CSDs, bottled water, fruit juice and sugar.25 26 36 43 44
The own-price elasticities for SSBs, in other studies, ranged between –0.548 and –2.1,26 while those for bottled water were higher, in absolute terms, ranging between –1.226 and –3.2.44 The own-price elasticity for sugar ranged between –1.143 and –2.4.25 Income elasticities for SSBs ranged between 0.45 and 1.20.36 Our findings for the own-price elasticity of CSDs are similar to the own-price elasticity for soft drinks in Guatemala26 and India.45 Our income elasticity estimates are consistent with findings from Ecuador,36 and suggest that the demand for these goods increases as income increases.
There are important policy implications of knowing how demand changes in response to an increase in SSBs prices. The most obvious one is the impact of imposing an SSB tax. In the past decade, a number of countries have implemented excise taxes on SSBs, with an explicit aim of improving public health.16
Mexico implemented an SSB tax of 1 peso (US0.07) per litre from 1 January 2014.50 Findings show that during the first year of the tax, the average SSB monthly purchases were 6% lower than would have been expected without the tax. In South Africa, the mean daily per capita SSB purchases fell by 5.2% from before to after the SSB tax announcement, and fell by a further 9.9% after the implementation of the tax.15 In the Philippines, SSB purchases fell by 8.7% in response to a tax-induced increase in SSB prices of between 16.6% and 20.6%.51
To the best of our knowledge, this is the first study to estimate SSB price and income elasticities for Nigeria. The results from this study could assist policy makers to estimate the likely impact of a tax-induced price change on the demand for SSBs.
As an example of how our results could be used to inform tax policy, consider the likely impact of the SSB tax of 10 naira per litre. Assuming that it will be fully passed through, the tax would increase the average retail price of CSDs by 4.3%, from 230 naira per litre31 to 240 naira per litre. Given a price elasticity of –0.76, this price increase would be expected to reduce the demand for CSDs by 3.3%.
One can estimate the size of the CSD market as follows: there are 41 million households in Nigeria, of which 20.1% consumed CSDs.31 The average consuming household consumed 1.4 L per week.31 Based on these numbers, the annual consumption of CSDs (pretax) is close to 605 million litres (41.4 million×20.1%×1.4 L×52 weeks). A 3.3% decrease in consumption because of the 10 naira per litre excise tax is expected to reduce consumption to 585 million litres and raise 5.8 billion naira in excise tax revenue.
Applying the same principle to malt drinks and chocolate powder suggests that the excise tax will raise a total of 19 billion naira. For chocolate powder, the excise tax revenue depends on the quantity of powder dissolved in water or other beverage. The analysis suggests that the excise tax revenue collected through this source varies between 10 billion naira (if 15 g of chocolate powder are used per 200 mL of beverage), and 29 billion naira (if 5 g of chocolate powder are used per 200 mL of beverage). Should the Nigerian government decide to follow a more stringent excise tax regime by raising the excise tax above the current rate, the decrease in consumption and the increase in tax revenue would be amplified. The SSB industry’s response to the excise tax increases can enhance the public health effect (eg, if the industry overshifts the tax, ie, increases the retail price by more than the tax increase), or reduce the public health effect (eg, if the industry undershifts the tax, ie, increases the retail price by less than the tax increase). The industry could reduce portion sizes without decreasing the retail price (shrinkflation). A well-considered SSB tax structure can have positive public health consequences, not just through the demand side (ie, demand reduction), but also through the supply side (eg, reduced sugar content through product reformulation) as was the case in Portugal, South Africa and the UK, where the SSB taxes are based on the sugar content.16
For an SSB tax to be fully effective in curbing the harmful effects of sugar-dense diets, the tax needs to be widened to include other sugar-laden beverages.16 The WHO recommends that SSB taxes be applied to all SSB categories including powders, concentrates or syrups used to make SSBs, in order to avoid the undesirable substitution of SSB consumption.16 Nigeria should also consider taxing powders used to make SSBs, such as chocolate powders, as is done in the Philippines. The Philippines has a tiered sugar tax of 6 pesos per litre (US$0.12) for drinks containing sugar and artificial sweeteners, and 12 pesos per litre (US$0.24) for drinks containing high-fructose corn syrup.16 Beverages taxed include sweetened juice, sweetened tea, CSDs, flavoured water, energy and sports drinks, powdered drinks, cereal and grain beverages, and sugar-sweetened non-alcoholic beverages.16
We note several limitations to this study. We do not report on all SSBs, only those reported in the NLSS 2018/2019 and consumed by households in the 7 days preceding data collection. Since the NLSS 2018/2019 data does not distinguish between sugar-sweetened and unsweetened beverages, we did not include juices and other SSBs in the study. The results are subject to self-reporting and recall biases as well as omitted variable bias due to lack of data on other factors affecting SSB demand. The risk of omitted variable bias is inherent in any econometric estimation. For example, the fact that there has been rapid growth in fast food outlets in Nigeria,4 and that SSB consumption is positively associated with visits to fast food outlets,52 could be a source of omitted variable bias, as such data was not available in the NLSS 2018/2019.
Lastly, we were unable to stratify our sample to estimate elasticities by income groups. Findings from other studies on SSB price elasticities have found that lower-income groups are more price sensitive, which reduces tax regressivity concerns.36 43 44 50 53
Despite these limitations, the evidence presented suggests that tax increases, which increase the price of SSBs, could reduce the demand for CSDs, malt drinks and chocolate powder, which, in turn, would curb SSB-attributable NCDs while simultaneously increasing SSB tax revenue. The WHO recommends that SSB taxes should apply to all categories of SSBs (including sweetened dairy drinks, powders, concentrates or syrups) but should exclude bottled or sachet water.16 Countries considering SSB taxes should not hesitate to do so, and should tax all categories of SSBs, as evidence from this study suggests that SSB taxes are effective at curbing SSB demand.
Data availability statement
Data are available in a public, open access repository. Data are available in a public, open-access repository. Data are available on request. The publicly available data can be accessed from: https://microdata.worldbank.org/index.php/catalog/3827/get-microdata.
Ethics statements
Patient consent for publication
Ethics approval
Not required. This study does not involve human participants.
Acknowledgments
The authors thank Nicole Vellios (Research Unit on the Economics of Excisable Products, University of Cape Town), and Alfred Kechia Mukong (University of Namibia) for their comments, and Guillermo Paraje (Business School - Escuela de Negocios) for his advice.
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors VD conducted data analysis and drafted the manuscript. CW provided guidance and review of data analysis and supported the drafting of the manuscript. VD is the guarantor. Both authors read and approved the final manuscript.
Funding Research Unit on the Economics of Excisable Products (REEP), School of Economics, University of Cape Town 109266-002 IDRC and the International Development Research Centre (IDRC), 109266-002 IDRC.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.