Article Text
PDF
Abstract
Introduction New Zealand’s Smokefree 2025 goal aims to greatly decrease the availability of tobacco. One option is to cease the sale of tobacco from convenience stores. However, tobacco companies and retail trade associations oppose this move and have argued that customers who purchase tobacco drive footfall and spend more than non-tobacco customers. The aim of this study is to test the validity of industry claims about the importance of tobacco to convenience stores.
Methods During November and December 2019, immediate postpurchase surveys were undertaken with customers on exit from a random sample of 100 convenience stores in two New Zealand cities. We estimated the mean number of items purchased, including tobacco and non-tobacco items, and mean expenditure on non-tobacco items.
Results Of the 3399 transactions recorded, 13.8% included tobacco, of which 8.3% comprised tobacco only and 5.5% included tobacco and non-tobacco items. The mean number of transactions containing both tobacco and non-tobacco items was 1.98, and 1.87 for those containing non-tobacco items only. Customers who purchased tobacco and non-tobacco items spent on average NZ$6.99 on non-tobacco items, whereas customers who purchased non-tobacco items only, spent on average NZ$5.07.
Conclusions Our results do not support claims that tobacco drives one-quarter of footfall into stores or that customers who purchase tobacco spend almost twice as much as non-tobacco customers. Combined purchases of tobacco and non-tobacco items constituted 5.5% of transactions; the impact on a store’s profitability of removing tobacco sales is unknown and could be the focus of future research.
- public policy
- advocacy
- economics
Data availability statement
Data are available upon reasonable request.
Statistics from Altmetric.com
Introduction
New Zealand (NZ) has experienced a reduction in smoking prevalence between 2006 and 2019; however, reductions have not occurred equitably.1 Smoking remains a key cause of health inequities; Māori are 2.73 times more likely to smoke than non-Māori.1 The current smoking prevalence for Māori is 34.0% compared with 14.2% for non-Māori.1 To meet the NZ government’s goal of reducing smoking rates to minimal levels by 2025 (usually interpreted as less than 5%),2 new measures are required to accelerate reductions in smoking prevalence. Strategies to achieve the 2025 goal include reducing tobacco availability,3–5 which was recommended as a core measure by the Māori Affairs Select Committee (MASC), who proposed the Smokefree 2025 goal.6 Reducing tobacco availability could help reduce health inequities, for example, by reducing the density of tobacco retailers in deprived areas where smoking is three times more prevalent.7 However, the NZ government is yet to implement measures to regulate tobacco supply and availability; opposition from retailers and associated tobacco interests is one barrier to this policy development.8
Tobacco companies are widely considered to influence small independent retailers by fostering relationships,9–11 offering incentives to stock and promote tobacco in-store,11–13 encouraging retailers to lobby against government policy,14 and by using marketing strategies to over-assert the importance of tobacco products and footfall (number of people entering a store) to retailers, such as assertions made in retail trade publications.14
Adult smokers in NZ primarily source their tobacco from small convenience stores, known as ‘dairies’.15 Although tobacco is a low-profit product,11 16 17 retailers claim they rely on the footfall and revenue generated from tobacco sales and the incidental non-tobacco transactions by customers who buy tobacco.16 17 Tobacco manufacturers and retailer associations have made statements arguing that tobacco is a crucial product for convenience store businesses; examples of specific claims that have been made include: (1) that tobacco products promote over one-quarter of footfall into shops18 and (2) that smokers spend almost twice as much on non-tobacco items as the average convenience shopper.14 19 Retailers’ concerns about the potential financial impact of tobacco availability reduction policies could undermine efforts to introduce policies that reduce tobacco availability.
There is a relative lack of NZ-related literature on retailers’ attitudes to selling tobacco;9 10 17 to date, only one study has specifically examined the characteristics of tobacco transactions in convenience stores.20 Robertson et al used a postpurchase survey to describe the patterns and prevalence of tobacco transactions in 20 convenience stores in Dunedin, NZ.20 That study found 14% of transactions included tobacco, 5% included both tobacco and non-tobacco items, and 9% were tobacco-only transactions. However, customers who purchased non-tobacco products only spent more on average than tobacco purchasers on non-tobacco-products (NZ$6.85 vs NZ$5.11).
Robertson et al’s findings are consistent with an earlier American study that explored the characteristics and prevalence of tobacco transactions at corner stores situated in low socioeconomic areas.21 Postpurchase surveys conducted with store patrons revealed 13% of transactions included tobacco, of which 8% were tobacco-only transactions and 5% included both tobacco and non-tobacco products. Irrespective of whether tobacco was purchased, researchers found no significant difference in the amount customers spent on non-tobacco products.
This study replicated and extended Robertson et al’s study,20 and tested the validity of claims made about tobacco footfall in convenience stores. Specifically, that customers who purchase tobacco drive footfall into stores and spend more than non-tobacco customers. The study drew on a stratified sample in two urban cities to provide more robust estimates of the proportion of tobacco transactions made at convenience stores in urban areas of NZ, a country with a progressive tobacco policy environment.
Methods
Immediate postpurchase surveys were undertaken with customers after they exited a sample of convenience stores in two NZ cities. This methodology minimises recall bias associated with self-reported transactions and has previously been used in research investigating factors associated with tobacco transactions.20–24
Sampling
The sampling units were convenience stores (referred to as ‘stores’ in the remainder of the paper), which are defined as small businesses that sell primarily food, beverages and a limited range of household goods. In NZ, these stores are not permitted to sell alcohol.25 A sample size of 100 stores (50 stores per urban city) provided a precision estimate of the proportion of tobacco and non-tobacco transactions.20 Each store was considered a cluster, as multiple customers were surveyed at each store. Using a random number generator, we selected eligible stores at random to a total of 50 in each city, Auckland (out of 756) and Wellington (out of 233), from a database developed previously by the researchers.26 Auckland is NZ’s largest city with a population of approximately 1.5 million, and Wellington is the capital of NZ with a population of approximately 200 000. Of the stores in the database, 5 stores were deemed ineligible in Wellington (3 were found to have closed since the database was developed, and 2 were identified as selling alcohol) and 15 stores were ineligible in Auckland (6 were closed or under renovation, 5 were unable to be located online, 3 were defined as a supermarket and 1 was a duplicate listing). The 50 randomly selected stores in each city were then mapped using ArcGIS,27 and grouped into local geographic areas for data collection. An additional 50 replacement stores for each city were identified using the same approach and were used when data collection was not able to be undertaken at a selected store (eg, it had closed, retailers declined to take part or did not sell tobacco). Replacement stores were grouped according to their location within each city in order to minimise travel and time demands on data collectors.
Data collection
Interviews took place outside each store at three 60 min intervals: during the morning peak traffic hour (07:00–11:00), during lunchtime (11:00–15:00) and during the late afternoon peak traffic period (15:00–19:00). Based on our earlier work, we anticipated completing a minimum of 10 surveys during each 60 min interval. Data collection took place in both cities from Monday to Friday over 5 weeks in November and December 2019.
On arrival, a trained interviewer entered each convenience store to assess eligibility. Stores were excluded if tobacco was not sold or if alcohol was sold; in these cases, the interviewer moved on to the replacement store in the geographic area. Interviewers then introduced themselves to the retailer present in the store and explained the study purpose, how the data would be used, the length of time they would be standing outside the store, and the planned day and time of subsequent visits to the store. An information sheet was provided to the retailer. Informed consent was not sought from the retailer as we were neither seeking personal information from them nor conducting any research on their premises. However, if any retailers objected to the research being conducted outside their store, the outlet was excluded, and the interviewer moved on to the replacement store. Stores that met eligibility criteria were then assessed for any unusual features (eg, whether they specialised in ice creams or other takeaway foods, or sold lottery tickets and/or instant scratch-cards). The NZDep2013 index was used to measure socioeconomic deprivation at the meshblock level for each tobacco retailer.28 A meshblock is the smallest geographic unit for which statistical data are reported in NZ, and varies in size from part of a city block to large areas of rural land. This index combines nine variables from the 2013 census that reflect eight dimensions of deprivation. The NZDep2013 scale provides an ordinal score from 1 to 10, where 1 represents the area with the least deprived score and 10 the areas with the most deprived score. We recoded the scores as low (1–3), medium (4–7) and high deprivation (8–10).
After informing retailers about the study, interviewers stood outside the exit and, at the start of the 60 min interval, invited the first person to exit the store to take part in the survey. Interviewers then interviewed the next person to come out of the store after they had finished the first interview, and so on, until the 60 min time period had lapsed. All customers who made a purchase from a convenience store were eligible for inclusion as a survey participant. Young people who appeared under the age of 15 years were asked for their age. Those aged 12 years and older, for whom we had ethical approval to interview, were included in the study; however, unaccompanied children aged under 12 years old were excluded. Individuals who appeared to have a diminished ability for understanding information about the study and providing informed consent (eg, individuals with the limited English language or those who appeared to have cognitive impairments) were also excluded. Details of ineligible participants were recorded together with the number of people who declined to take part. Those who exited the store as an interview was in progress were noted.
Participants were asked how many items they purchased, whether they purchased tobacco and what non-tobacco items, including smoking paraphernalia (ie, lighters, matches, papers and filters), they bought. The total amounts spent in the transaction and on tobacco (where a tobacco product was purchased) were also requested. Data collectors noted if participants produced a receipt or relied on a recollection of the transaction total spend. These data were recorded on an iPad using the Qualtrics survey platform. The University of Otago Human Ethics Committee approved this project (ref 17/132). We consulted with the Ngāi Tahu Research Consultation Committee to assess the importance of the project to Māori.
Analysis
For this study, an item is an individual product (tobacco or non-tobacco) included in a transaction. A transaction refers to an individual customer who purchases one or more items (which may include tobacco and/or non-tobacco items).
We produced summary statistics for (1) the average number of items per transaction, (2) the average number of non-tobacco items per transaction and (3) the average spend on non-tobacco items per transaction. In addition, we examined the items purchased in the following subgroups: (a) transactions of non-tobacco items only, (b) transactions of tobacco items, (c) transactions of customers who purchased tobacco items only (a subset of b), and (d) transactions including both non-tobacco items and tobacco (also a subset of b). All analyses were conducted using Stata V.16.29
For each subgroup, its proportion of the full sample was calculated and the CIs were produced using binomial exact intervals.
The average number of tobacco and non-tobacco items per transaction (points (1) and (2) mentioned earlier) and CIs were calculated using Poisson mixed models (with robust standard errors) with a random effect on the store identification code to allow for the clustering in the sample. Only an estimate of the mean count was required; therefore, the model included no predictors but allowed for clustering. This allowed us to produce a mean for both items per transaction and non-tobacco items per transaction. Poisson regression was required because the data were counts of items purchased.
The mean and CI for the non-tobacco spend were calculated using a linear mixed model (with robust standard errors). Again, to estimate the mean, while allowing for clustering, the model included a random effect by store but had no predictors. As customers tend to spend small amounts in convenience stores, the data on spending were highly skewed and the assumption of normally distributed errors was violated. Consequently, the spend was log-transformed. Therefore, after fitting the model, the results were back-transformed to give the geometric mean and the corresponding CI.
Results
Data were collected at 104 stores (table 1) including 72 in the original sample and 32 were a replacement. The response rate was 69%. The reasons that full data were not collected at stores (whether those originally selected or replacement stores) were: retailers declined to participate (11), store closed (7), retailer sold alcohol (6), staff member unable to give consent to participate (4), store unable to be located (2) and store did not sell tobacco (2). Where data were not able to be collected for the full three data collection periods for six selected stores, the remaining data collection occurred at replacement stores.
Data collection and characteristics of convenience stores
The majority of stores (41%) were located in high deprivation areas, 40% in medium deprivation areas, and 18% in low deprivation areas (table 1).
There were 4936 people who exited a store as the study was being undertaken. Of these, 1211 did not take part in the study as they were either unable to provide consent, were under the age of 12 years, exited the store as the interviewer was undertaking another survey or declined to take part without reason being given. A further 326 people consented to take part but were excluded from the final analyses because they did not spend anything in the store they visited. The number of transactions recorded were 3399 over the data collection period, including 1530 in Auckland and 1869 in Wellington (table 2). Of these transactions, 2.3% of people provided a receipt of their expenditure, 89.8% provided a confident recollection of their expenditure and 7.5% were based on an estimate. Overall, 13.8% of transactions included tobacco, of which 5.5% included non-tobacco items and 8.3% comprised tobacco only (table 2). Results for Auckland and Wellington are presented separately in the online supplemental files 1 and 2, respectively.
Supplemental material
Supplemental material
Convenience store transactions—full sample and non-tobacco-only sample
Overall, the mean number of non-tobacco items purchased for the full sample of customers who purchased items from these stores was 1.71 (table 2). The mean number of non-tobacco items for customers who did not purchase tobacco was 1.87 and the mean number of non-tobacco items for the group of customers who purchased both tobacco and non-tobacco items was 1.98 (table 2).
In terms of expenditure, customers who purchased tobacco and non-tobacco items spent on average NZ$6.99 on non-tobacco items, whereas customers who purchased non-tobacco items only, spent on average NZ$5.07 (table 2).
The most common purchases from stores were ready-to-eat food such as takeaways, fruit, snacks, ice creams, pies, lollies, chewing gum and muffins. Tobacco (including tailor-made cigarettes and RYO) was the third most common item purchased (13.8% of transactions overall). Compared with soft drink (40.1%) and ready-to-eat food (42.1%), tobacco items were much less common, although there were more tobacco items purchased (13.8%) than phone cards (4.8%), which were fourth in the rankings. Smoking paraphernalia was purchased in 2.8% of transactions.
Discussion
Our study extended the original study undertaken in Dunedin to explore the validity of industry claims made about tobacco footfall. One example of such a claim maintained that tobacco products promote one-quarter of footfall into shops.18 This claim was not supported by the findings from our study, which found that 14% of transactions made in convenience stores included tobacco items. These results are also consistent with Robertson et al,20 who found that 14% of transactions in convenience stores included tobacco items, and with Lawman et al’s study of corner stores in a low-income area of Philadelphia, that similarly found 13% of transactions included tobacco items.21
In contrast to Robertson et al 20 and Lawman et al,21 customers purchasing tobacco in this study spent more per transaction on non-tobacco items than customers who only purchased non-tobacco items. The difference in the average spend per transaction in the full sample between those who bought tobacco and those who did not was NZ$1.92. A further claim made by tobacco companies and retail trade association advocates that customers who purchase tobacco items spend almost twice as much on non-tobacco items compared with customers purchasing non-tobacco items only.14 19 However, our study did not find evidence to support the claim.
Combined purchases of tobacco with non-tobacco items constituted 5.5% of transactions investigated. These results are also consistent with Robertson et al 20 who found that 5% of transactions included both tobacco and non-tobacco items; likewise, Lawman et al 21 found 5% of transactions included tobacco and non-tobacco items in corner stores in Philadelphia. It is difficult to determine with any certainty the impact on a store’s profitability of removing 5.5% of tobacco purchases without knowing the mark-up on tobacco and non-tobacco items. Tobacco is known as a low-profit product11 14 16 17 30 and profit margins can vary between products and stores.16 The mark-up on non-tobacco products is considerably higher compared with tobacco products, with UK data reporting 24% profit margins compared with around 4%–6% for tobacco products.14 Future research could consider the impact on convenience store profitability of ceasing sales of tobacco, including any loss from tobacco sales, as well as potential loss on accompanying sales of non-tobacco items.
Although retail reduction measures are yet to be implemented in NZ,6 reducing tobacco availability is a key strategy in the Achieving Smokefree Aotearoa Action Plan, developed in consultation with the wider tobacco control sector.31 Several policy options have been proposed,32 including allowing only specific retailers to sell tobacco. The policy option chosen could have differing effects on small retailers, such as convenience stores. For example, a policy disallowing tobacco sales in convenience stores but allowing tobacco sales in stores that sold competing non-tobacco goods will have an impact on the revenue and profitability of convenience stores. In contrast, a policy that allowed tobacco sales at tobacco-only stores restricted to people 18 years and older, and where competing non-tobacco products could not be sold, may have less impact on convenience stores. This would result in a reduction in sales of tobacco, but not on other products.
Retailers who sell tobacco are not completely resistant to tobacco retail reduction policies,9 33 34 and in the absence of regulatory policies, a small proportion of tobacco retailers have voluntarily stopped selling tobacco.35–38 In many cases, the main reason reported for ceasing sales of tobacco includes low-profit margins or declining sales volumes.36–38 One of the few NZ studies reporting on the views of convenience store retailers who do not sell tobacco noted that the slight initial downturn was worth it or noted that increased community support strengthened their business.39 Relying on retailers to voluntarily stop selling tobacco is unlikely to reduce tobacco availability sufficiently to achieve the Smokefree 2025 goal. However, lessons from these businesses provide an opportunity to develop strategies that reflect their experiences.
Tobacco retail reduction policies need to assist small retailers to divest themselves of tobacco and transition to other higher-profit margin products. Other high volume items identified in this study include ready-to-eat products, which may have higher-profit margins than tobacco. However, if these products are high in fat, sugar, or salt then they may present additional health challenges and give rise to ‘food swamp’ communities (neighbourhoods saturated with unhealthy food, beverages, tobacco and alcohol).40 To avoid replacing one risk product with another, policy-makers need to adopt a broad focus on well-being and healthy environments. International initiatives, such as an intervention transforming neighbourhood corner stores into healthy food retailers undertaken in San Francisco, illustrate the potential value of alliances between small businesses and community health coalitions.41 Similar programmes that aim to increase access to healthy food and decrease exposure to unhealthy products, such as tobacco, could improve health equity and increase profit margins for retailers.
The current global pandemic associated with COVID-19 has already had a substantial economic impact on the retail sector that is likely to worsen.42 43 However, the pandemic offers opportunities ‘to filter our perception of what is possible through this lens’ and place a new priority on well-being.44 45 Whether the pandemic translates to an opportunity to fundamentally change the tobacco retail environment, in any country, including NZ, is yet to be seen. Espiner and colleagues highlight that other public health interventions implemented with the same focus and dedication as we have seen for COVID-19 could have widespread positive health implications for a large proportion of the population and, in particular, for indigenous populations.44 One of the intervention examples the authors suggest is the implementation of tobacco supply reduction policies. Unfortunately, it may be that the concerns about the viability of all retailers mean that no changes will be made, as any small contribution to profitability is seen as sacrosanct, no matter how harmful. Or, there may be more opportunities to gain traction to reduce the tobacco purchase environment, packaged up with the current high health literacy, and interest in public health.
The strength of our study is the large sample size from two major cities, which included 3399 transactions across 104 stores and three different periods of the day. Random sampling was used to select convenience stores in the two cities, and the response rate was high (69%) with data collected from 104 stores. Data entry was significantly reduced and streamlined by using the electronic software, and with the triangulation of data, using the location, date and time stamps, correction of any miscoding, such as store identity number or data collection phase, or other errors was simple.
Excluding children under the age of 12 years may have overestimated the proportion of transactions that included tobacco. The generalisability of these findings to NZ provinces is unknown; future research could focus on these non-urban areas. Our study relied on customers reporting the amount they spent in the store, a method that has been used in other studies.20–24 The use of immediate postpurchase surveys would be expected to reduce the occurrence of recall bias and estimates would not have differed based on whether tobacco was purchased. However, because of reliance on self-reported purchase data, more analyses could not be made into the impact of policies on these businesses. We were also unable to assess how these convenience stores would be impacted financially by tobacco retail reduction policies.
Our study did not differentiate among customers exiting stores in terms of demographic characteristics or smoking status. Postpurchase intercept surveys were used in a recently published US study that examined purchases made in gas station–convenience stores by 18–25 year olds who were susceptible to, or who used, tobacco. The study found that of the 33% of participants who purchased tobacco, 34% purchased food, 8% purchased lottery tickets and 8% purchased something not listed.24 These estimates are higher than our findings, and those of Robertson et al 20 and Lawman et al,21 which did not restrict their sample to youth and young adults who had used or were susceptible to tobacco use. Future research should consider identifying particular groups of the population that may be at greater risk, such as young adults, a group in which substantial smoking initiation occurs.46
Conclusion
Our study found that tobacco constituted 14% of observed transactions in convenience stores which does not support claims that tobacco drives footfall into stores. Our study also found that people who purchased non-tobacco items, when buying tobacco, spent on an average NZ$1.92 more per transaction on non-tobacco items than those purchasing non-tobacco items alone. These results do not provide evidence to support the claim that customers who purchase tobacco spend almost twice as much as customers who only purchase non-tobacco items. Combined purchases of tobacco with non-tobacco items constituted 5.5% of transactions; the impact on a store’s profitability of removing these tobacco sales is unknown and could be the focus of future research.
What this paper adds
What is already known on this subject
Tobacco retailer groups suggest tobacco products promote over one-quarter of footfall into shops, that is, the tobacco footfall argument.
Tobacco companies suggest that smokers spend almost twice as much on non-tobacco items as the average convenience shopper.
What important gaps in knowledge exist on this topic
Only two published studies to date have examined the tobacco footfall argument, including one small New Zealand study.
What this paper adds
This study replicated and extended the original New Zealand study to test the validity of claims made about tobacco footfall with a larger sample in two urban cities.
Our study does not support the claim that tobacco drives one-quarter of footfall into stores.
Our study does not provide evidence to support the claim that customers who purchase tobacco, spend almost twice as much on non-tobacco items as customers who only purchase non-tobacco items.
Data availability statement
Data are available upon reasonable request.
Ethics statements
Patient consent for publication
Ethics approval
The University of Otago Human Ethics Committee approved this project (ref 17/132).
Acknowledgments
The authors wish to thank Dr Crile Doscher, Lincoln University, for mapping the convenience stores in this study, and Dr Philip Gendell, University of Otago, for comments on the study design and draft manuscript. They would also like to extend our gratitude to the tobacco retailers who allowed them to collect data outside their store, and the many customers who took the time to complete the survey.
References
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 LM, CC, JH, TS and LR conceptualised the project. LM obtained the funding. LM, RQ, CC, LR and M-LB designed the study protocol. LM, SW and M-LB managed the data collection process. RQ developed the Qualtrics survey and managed the data. NV and LT obtained the data. CC undertook sampling, conducted the analysis, interpreted data and prepared the results. SW conducted the literature review. LM drafted the manuscript. All authors provided feedback on subsequent drafts of the manuscript. All authors have seen and approved the final version.
Funding This study was funded by Lottery Health Research (R-LHR-2019–1 01 847).
Competing interests None declared.
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.