Sugar-sweetened beverage price elasticities in a hypothetical convenience store
Introduction
One policy-active area of public health is sugar-sweetened beverage (SSB) purchase and consumption. SSBs are generally described as non-alcoholic water-based beverages with added sugar and may include cordials, fruit drinks, flavoured milks, sports drinks, energy drinks, vitamin waters, sweetened ice teas, as well as non-diet sodas or ‘soft drinks’. These products offer minimal nutritional benefits and their consumption is associated with obesity, Type 2 Diabetes Mellitus and dental decay (Malik et al., 2010; Trumbo and Rivers, 2014). Purchasing decisions for these products vary within and across consumer groups, including by age, gender, income, frequency of purchase, and race/ethnicity (Australian Bureau of Statistics, 2014; Etilé and Sharma, 2015; Han and Powell, 2011, 2013).
Over recent years there has been an acceleration in the number of governments considering or implementing health-motivated beverage pricing changes, including SSB taxes (Backholer et al., 2016a), and health-related food taxes (Cornelsen et al., 2014). Recently the World Health Organisation (WHO) urged governments to adopt fiscal policies to improve the healthiness of food and beverage intake (World Health Organisation, 2015; World Health Organization, 2016). Health-motivated SSB taxes are designed to discourage SSB consumption to reduce energy and added sugar intakes (World Health Organisation, 2015; World Health Organization, 2016).
In this paper, we focus on four particular gaps in the literature, which need to be addressed to inform evidence-based policy developments: (i) estimates of beverage price elasticities over a wide range of price levels; (ii) price elasticity estimates for policy-relevant subgroups; (iii) investigation of clustering of consumer preferences and behavioural responses to beverage pricing interventions; and (iv) examination of price elasticities in a convenience store setting.
“[Price] elasticities give the percentage change in demand for good per (marginal) percentage change in the price of good (Mas-Colell et al., 1995). This can be the change in demand for a beverage in response to changes in its price (own-price elasticity (OPE)) or change in demand for a beverage in response to change in the price of another beverage (cross-price elasticity (CPE)). Accurate price elasticity estimates are required to predict both changes in purchasing, and the effect on health-related outcomes such as weight gain, which can be extrapolated to estimate population level health effects such as obesity rates (Veerman et al., 2016). Own and cross-price elasticity estimates for policy-relevant population subgroups (e.g. based on age, gender, income and frequency of SSB consumption) would improve predictions of differential intervention effects and thereby assist in determining who is likely to be most responsive to pricing interventions (Nghiem et al., 2013). Further, better understanding the clustering of consumer preferences and behavioural responses to beverage pricing interventions could improve predictions of whose health is likely to benefit most from pricing interventions.
Natural experiments provide important and externally valid insights into the effects of a price shock such as an SSB tax on entire populations (Craig et al., 2012). A growing number of real-world SSB tax evaluations suggest that SSB taxes are associated with decreases in SSB purchases (Caro et al., 2018; Colchero et al., 2016, 2017; Silver et al., 2017). However, natural experiments do not allow estimation of price elasticities over a range of prices and therefore cannot be used to calculate price elasticities and cannot be used for testing yet-to-be implemented policy proposals, such as combined SSB taxes and non-SSB subsidies (Craig et al., 2012).
To date, price elasticity studies have predominantly used historical purchasing data, based on time series, household surveys, or scanner panel data (Andreyeva et al., 2010 Feb; Escobar et al., 2013; Harding and Lovenheim, 2017; Powell et al., 2013; Sharma et al., 2014). They have found a wide range of population beverage price elasticity estimates for different beverage groups. For example, the most recent review in 2014 found an OPE for non-diet soda of −1.25 (range −0.71 to −2.26). Elasticity estimates from historical purchasing data are subject to several issues, including: elicitation from a narrow range of price levels leading to extrapolation of elasticity estimates beyond the data price variation, reducing generalisability of the results (Nghiem et al., 2013); and exclusive focus on supermarket settings. The focus to date on purchases from supermarkets overlooks approximately a quarter of off-trade (non-restaurant) purchases of ready-to-drink beverages from non-supermarket sources such as small grocery stores, convenience stores and vending machines (Euromonitor International, 2015).
Beverages in convenience store settings generally have a higher unit price per volume than supermarket settings (Euromonitor International, 2015), and often have smaller individual beverage sizes dominating offerings. Further, supermarket estimates may be reflective of ‘stockpiling’ or ‘forward-purchasing’ behaviour in which consumers purchase more of a product when it is less expensive, in expectation that they will buy less during more expensive periods (Mela et al., 1998). By contrast, on-the-go beverage purchasing, such as in convenience stores, may plausibly be associated with greater impulse purchasing, but this has not been studied to date. Hence differences in alternatives available, alternative attributes (including price and volume), and choice purchasing context (‘task’), may all affect preferences, price elasticities, and purchasing patterns (Yale and Venkatesh, 1986). This may therefore also affect response to potential pricing policies, warranting further investigation in these settings.
Moreover, there have been very few studies on drivers of heterogeneous price elasticities of demand or response to SSB pricing policies amongst population subgroups. Limited evidence from historical purchasing data suggests mixed results on relative SSB price sensitivity across income groups (Backholer et al., 2016b), and that more frequent SSB consumers may have lower SSB price sensitivity than less frequent consumers (Etilé and Sharma, 2015; Finkelstein et al., 2013), particularly among older consumers (Dubois et al., 2017). The scarcity of micro-level data limits both this subgroup analysis and the examination of individual-level price responsiveness. Finally, to our knowledge, no other work has identified latent-class derived consumer group reactiveness to beverage price changes until now.
Discrete choice experiments (DCEs) offer a number of advantages in the exploration of consumer reactiveness to public health interventions in a hypothetical choice scenario. This includes the ability to explore elasticities over a wider range of price levels than usually observed in market data and the avoidance of multicollinearity by design. There have been no published DCEs to date examining packaged beverage purchases, however there is emerging work in this field in children (Yang et al., 2016) and using somewhat related methods in adults (Zizzo et al., 2016). There is a related literature using online randomised control trials comparing responses of participant groups to hypothetical price increases on either both unhealthy food and beverages (Epstein et al., 2010; Nederkoorn et al., 2011), or on SSBs only (Bollard et al., 2016; Waterlander et al., 2014). These experiments have shown lower numbers of calories purchased when unhealthy food and/or beverages are subjected to price increases. While RCTs have strong causal properties, as all SSBs were subjected to the same price increase in each treatment scenario, these experiments have not allowed simulation of scenarios where different beverages are subject to price increases, or allowed estimation of substitution behaviour between beverage alternatives (i.e. accurate price elasticity estimation). Lastly, these works have exclusively been carried out in supermarket settings and have included limited population subgroup analyses, mainly due to small sample sizes.
In this study, our overarching purpose was to determine the effect of changing beverage prices on beverage purchases at an aggregate level and in different consumer groups. We therefore aimed to: (i) determine pre-packaged beverage price elasticities for the overall sample and for common policy-target consumer subgroups, including by age, gender, SSB consumption frequency and income level; and (ii) identify consumer subgroup clusters that are likely to be most reactive to beverage price changes.
Section snippets
Overview
To answer these research questions we undertook an online computer-based DCE with a nationally representative sample of the Australian general public. To address the first aim we use a mixed logit model which allows for accurate estimation of unrestricted substitution patterns (Revelt and Train, 1998), which is important for predicting the likely health outcomes of pricing policies, since health outcomes differ due to nutrient and energy content, both of which differ across beverage
Study population
Between May and June 2016, 1,008 eligible adults completed the DCE drawn from an online panel. The sample was representative of the Australian population in age, gender (Australian Bureau of Statistics, 2011) and income tertiles (Australian Bureau of Statistics, 2013) (Table 2). Median [Interquartile range] survey completion time was 16.5 [12.1, 24.2] minutes.
Mixed logit results for pre-specified consumer subgroups
Goodness of fit increased progressively from Model A to B to C (Table 3). Due to superior goodness-of-fit, Model C was used for
Discussion
In this study, we have advanced the literature on the generation of beverage OPEs and CPEs in a number of important respects. Using a DCE, we have been able to look at larger price variation than is usually found using historical purchasing data, thereby allowing generalization to policy-relevant magnitudes of beverage price changes. We have uniquely focused on beverage pricing responses within a convenience store setting. To our knowledge, we are the first to investigate clustering of consumer
Acknowledgements
This research was funded by a Monash University Faculty of Business and Economics Interdisciplinary Grant. The funder had no role in the collection, analysis or interpretation of data. MB is supported by an Australian Government Research Training Program Scholarship and Deakin University. AP and KB are supported by Deakin University. EL is supported by an Australian Research Council fellowship. The authors thank Peter Sivey for his feedback on the manuscript and Hong Il Yoo, Yuanyuan Gu, Gang
References (68)
- et al.
The effect of sugar-sweetened beverage price increases and educational messages on beverage purchasing behavior among adults
Appetite
(2018) - et al.
Implications of a sugar-sweetened beverage (SSB) tax when substitutions to non-beverage items are considered
J. Health Econ.
(2013) - et al.
Effect of food prices on the prevalence of obesity among young adults
Publ. Health
(2011) - et al.
Consumption patterns of sugar-sweetened beverages in the United States
J. Acad. Nutr. Diet.
(2013) - et al.
The effect of prices on nutrition: comparing the impact of product-and nutrient-specific taxes
J. Health Econ.
(2017) - et al.
Factors affecting food choices of older adults from high and low socioeconomic groups: a discrete choice experiment
Am. J. Clin. Nutr.
(2015) - et al.
High tax on high energy dense foods and its effects on the purchase of calories in a supermarket. An experiment
Appetite
(2011) A structural equation model of latent segmentation and product choice for cross-sectional revealed preference choice data
J. Retailing Consum. Serv.
(1994)EM algorithms for nonparametric estimation of mixing distributions
Journal of Choice Modelling
(2008)- et al.
Effects of a price increase on purchases of sugar sweetened beverages. Results from a randomized controlled trial
Appetite
(2014)
The impact of food prices on consumption: a systematic review of research on the price elasticity of demand for food
Am. J. Public Health
Australian demographic statistics, Jun 2016, Cat no. 3101.0
Household income and income distribution Australia, Cat no. 6523.0
Australian Health Survey: nutrition first results - foods and nutrients, 2011–12, Cat no. 4364.0.55.007
National Health Survey: first results, 2014-15, Cat no. 4364.0.55.001
Education and Work, Australia, May 2016, Cat no. 6227.0
Have we reached a tipping point for sugar-sweetened beverage taxes?
Publ. Health Nutr.
The impact of a tax on sugar-sweetened beverages according to socio-economic position: a systematic review of the evidence
Publ. Health Nutr.
Retailer-led sugar-sweetened beverage price increase reduces purchases in a hospital convenience store in Melbourne, Australia: a mixed methods evaluation
J. Acad. Nutr. Diet.
Effects of plain packaging, warning labels, and taxes on young people's predicted sugar-sweetened beverage preferences: an experimental study
Int. J. Behav. Nutr. Phys. Act.
Understanding heterogeneous preferences in random utility models: a latent class approach
Environ. Resour. Econ.
Overall and income specific effect on prevalence of overweight and obesity of 20% sugar sweetened drink tax in UK: econometric and comparative risk assessment modelling study
Br. Med. J.
Chile's 2014 sugar-sweetened beverage tax and changes in prices and purchases of sugar-sweetened beverages: An observational study in an urban environment
PLoS Med.
Pass-through of a tax on sugar-sweetened beverages at the Philadelphia International Airport
J. Am. Med. Assoc.
Beverage purchases from stores in Mexico under the excise tax on sugar sweetened beverages: observational study
Br. Med. J.
In Mexico, evidence of sustained consumer response two years after implementing a sugar-sweetened beverage tax
Health Aff.
What happens to patterns of food consumption when food prices change? Evidence from a systematic review and meta-analysis of food price elasticities globally
Health Econ.
Using natural experiments to evaluate population health interventions: new Medical Research Council guidance
J. Epidemiol. Community Health
How well targeted are soda taxes?
The influence of taxes and subsidies on energy purchased in an experimental purchasing study
Psychol. Sci.
Evidence that a tax on sugar sweetened beverages reduces the obesity rate: a meta-analysis
BMC Public Health
Do high consumers of sugar-sweetened beverages respond differently to price changes? A finite mixture IV-Tobit approach
Health Econ.
Soft Drinks: Euromonitor from trade sources/national statistics
Preferences for new and existing contraceptive products
Health Econ.
Cited by (9)
Optimisation of natural sweeteners for sugar reduction in chocolate flavoured milk and their impact on sensory attributes
2021, International Dairy JournalCitation Excerpt :Globally, consumption of added sugar has tripled in the last 50 years (Lustig, Schmidt, & Brindis, 2012). Major contributors to consumption of added sugar are sugar-sweetened beverages which include soft drinks (e.g., soda or pop), fruit and sports drinks, tea, coffee, energy drinks, sweetened milk or milk alternatives, as well as other beverages to which sugar (mainly sucrose, fructose or high fructose corn syrup) has been added (Blake, Lancsar, Peeters, & Backholer, 2019; Cleghorn et al., 2019; Falbe, Thompson, Patel, & Madsen, 2019). Milk is the principal dairy product consumed by children worldwide.
Can incentive-compatibility reduce hypothetical bias in smokers’ experimental choice behavior? A randomized discrete choice experiment
2020, Journal of Choice ModellingCitation Excerpt :Some tobacco studies have applied ex-post techniques, namely econometric calibration using revealed preference data, to correct for hypothetical bias (Kenkel et al., 2020; Buckell and Hess, 2019). Calibration of market shares has also been applied in studies of sugar-sweetened beverages (Blake et al., 2019). Other available approaches have not yet been tested in purchasing-based DCEs, and may have value to researchers.
An integrated modelling approach examining the influence of goals, habit and learning on choice using visual attention data
2020, Journal of Business ResearchCitation Excerpt :The relationship between visual attendance and participant demographics, beverage preference and choice characteristics was explored using an eye-tracked DCE. Details of the DCE without the addition of eye-tracking data have been published (Blake et al., 2018; 2019) which report on the DCE applied to different, larger samples than used in the eye tracking dataset used in this current study. Briefly, the primary purpose of the DCE was to explore heterogeneity in consumer beverage preferences and price responsiveness over key socioeconomic characteristics including income levels and usual SSB consumption frequency.
Modeled health economic and equity impact on dental caries and health outcomes from a 20% sugar sweetened beverages tax in Australia
2023, Health Economics (United Kingdom)Comparison of a full and partial choice set design in a labeled discrete choice experiment
2023, Health Economics (United Kingdom)