Introduction

Decreasing tobacco retail availability could reduce youth smoking initiation and facilitate cessation by increasing the ‘full cost’ of cigarettes (ie, travel time and associated costs), reducing environmental smoking cues and denormalising tobacco.1–5

However, opposition from retailers may impede adoption of tobacco retail reduction policies, particularly as owners of convenience stores (or ‘dairies’ in New Zealand (NZ)) claim to rely on tobacco sales, despite the low profit margins (typically <10%) on tobacco products.6–9 Some small retailers and tobacco industry-funded retail associations claim tobacco constitutes one-third to two-thirds of convenience store sales.8 10 They also assert that tobacco increases footfall, and that customers who purchase tobacco tend to buy other higher margin non-tobacco items, thus increasing overall turnover and profit.7 Tobacco manufacturers cultivate relationships with small retailers,7 8 11 offer them incentives to stock tobacco12–14 and promote tobacco as a core item in their product range.15 One Imperial Tobacco trade advertisement claimed, ‘Smokers don’t just buy tobacco products. Their basket spend is more than twice as much as the average convenience shopper.’ 15 This marketing may lead retailers to overestimate the economic importance of tobacco.

To our knowledge, only one published study has objectively assessed the importance of tobacco to small retailers. A postpurchase survey in Philadelphia (USA) found only 13% of transactions included tobacco; most were tobacco-only purchases (8%) and only 5% included both tobacco and non-tobacco items.16 Further, there was no difference in expenditure on non-tobacco items when purchased with or without tobacco.16 While these results are inconsistent with tobacco industry arguments that tobacco is crucial to small retailers, replications of this study are necessary, particularly in different regulatory environments. Therefore, we examined the prevalence and characteristics of tobacco purchases at convenience stores in Dunedin, NZ.

For context, Dunedin is a small urban area17 with a population of around 127 000 people.18 Eighty-eight per cent of the residents are identify as European and 8% as Māori (NZ’s indigenous people), compared with 74% and 15%, respectively, for NZ.19 The Southern region, where Dunedin is situated, has a smoking prevalence of 16.9% which is comparable with that of NZ as a whole (16.2%).20 Around 48% of NZ adult smokers usually purchase their tobacco from a convenience store.21 NZ has historically had a progressive tobacco control strategy, and the government has implemented annual 10% tax increases on tobacco since 201022 (excise tax now constitutes around 62% of tobacco’s retail price, meaning NZ is one of the most expensive countries in which to buy tobacco).23 Despite this, NZ has very few discount tobacco shops, and most retailers sell tobacco at the recommended retail price.22

Methods

The sampling units were convenience stores, defined as small businesses that sell primarily food, beverages and a limited range of household goods; in NZ, they are not permitted to sell alcohol.24 We selected 20 outlets (approximately one-third of Dunedin’s convenience stores): 13 from high-deprivation areas and 7 from mid-to-low-deprivation areas25 which approximates the wider distribution of outlets in the sampling frame (around 64% of Dunedin convenience stores are in high-deprivation neighbourhoods and 36% are in mid-to-low-deprivation neighbourhoods; which is very similar to the distribution in NZ as a whole).26 We purposefully sampled outlets geographically to ensure representation of suburbs, urban areas and commuting routes, and to avoid clustering of units within any particular area. Our final sample was drawn from 17 different neighbourhoods across Dunedin. Data were collected from each store at three 60 min intervals, in the morning (07:30–08:30 hours), lunchtime (11:30–12:30 hours) and late afternoon (15:00–16:00 hours), over a 4-week period from November to December 2017. We randomised the study week and day of week (Monday to Friday) for each data collection interval.

A trained interviewer informed each retailer about the study and noted any store features that could affect footfall (eg, postal services, lottery tickets available, extensive range of takeaway foods, special offers). The interviewer stood outside the store and invited each person exiting to participate in the survey; individuals were recorded as ‘missed’ if they were exiting the store as another survey was in progress, and thus could not be invited to participate. We excluded unaccompanied customers aged under 12 years and individuals with limited English. Participants were asked how much they had spent in the store, details of purchases (ie, size, quantity, brand) and the cost of tobacco, if purchased. As convenience stores do not typically provide receipts, the interviewer recorded whether the participant appeared certain of, or had approximated, their expenditure. Where expenditure data were missing completely, we used government tobacco returns data27 to assign a tobacco price, and verified prices of non-tobacco items through store visits.

We used descriptive statistics to determine the proportions (and 95% CIs) of transactions that (1) contained only non-tobacco items, (2) contained tobacco, (3) contained only tobacco and (4) contained both tobacco and non-tobacco items. These proportions were calculated for the whole sample, and for each individual outlet. We estimated the mean number of items purchased, the mean number of non-tobacco items purchased and mean expenditure on non-tobacco items. We defined smoking paraphernalia (ie, lighters, matches, papers and filters) as non-tobacco items.

Results

One retailer did not permit the interviewer to conduct research outside the store, giving a sample of 19 convenience stores. Of 954 potential participants invited to take part, 245 declined, giving a response rate of 74.3% (online supplementary file 1). We recorded a further 207 ‘missed’ individuals and 111 unaccompanied children under 12 years of age. We excluded 30 participants who made no purchase, giving a final participant sample of n=679. Eleven per cent of participants were recorded as being uncertain of the exact expenditure. Missing or ‘uncertain’ data were verified for all but 28 participants (where a lack of recorded detail precluded expenditure verification).

Of the 679 transactions, 14.0% contained tobacco (see table 1), 11.8% included tailor-made cigarettes and 2.2% included roll-your-own tobacco. Of the 95 transactions containing tobacco, the majority (64%) comprised tobacco only, whereas 36% contained tobacco and non-tobacco items. Transactions that included tobacco and one or more non-tobacco items represented 5% of all transactions.

The mean number of non-tobacco items purchased was 1.8 for the sample as a whole, 1.9 for transactions containing only non-tobacco items and 1.7 for transactions containing both tobacco and non-tobacco items (see table 1). Among the 34 transactions that included tobacco and non-tobacco items, the most common non-tobacco item was smoking paraphernalia (eg, matches, lighter, papers or filters; n=33 items), followed by takeaway and snack food (n=15 items) and soft drinks (n=12 items). The mean expenditure for transactions including tobacco was $33.74; people who purchased only tobacco spent on average $29.83, and those who purchased both tobacco and non-tobacco items spent $41.31. After excluding the cost of tobacco, people who purchased tobacco and non-tobacco items spent on average $5.11 on non-tobacco items, whereas people who purchased only non-tobacco items spent $6.85. People who bought both tobacco and non-tobacco items were more likely to have bought more than one packet of tobacco, compared with those who bought tobacco only (9% vs 5%, respectively).

The most common purchase was takeaway and snack food which made up 54% of the 1313 items across all transactions, followed by soft drinks (22%). Tobacco was the third most common item purchased (8%) and was purchased as frequently as milk, bread, eggs, other food and household items combined (8%).

There was considerable variation in the proportion of transactions that included tobacco across outlets (range 1%–45%; table 1). For example, in six retailers, tobacco constituted over a quarter of all transactions. This variation likely reflects differences in store characteristics. For instance, one outlet, where 45% of transactions contained tobacco, sold tobacco products at less than the recommended retail price. Generally, the prevalence of tobacco purchases was lower at stores that had a higher turnover of customers within the 60 min study intervals, than at lower turnover stores (table 1).

Discussion

Our study, among the first to objectively assess the prevalence and characteristics of tobacco purchases, found the vast majority (86%) of transactions at Dunedin convenience stores did not include tobacco. Of the 14% of purchases that did include tobacco, most (64%) were tobacco-only. Purchases of tobacco with non-tobacco items accounted for just 5% of transactions, and tobacco-purchasers spent less money on non-tobacco items than those who did not purchase tobacco. This finding may reflect the association between smoking and socioeconomic status28 and the fact that smokers may, on average, have less disposable income than non-smokers. These findings are almost identical to the results of an intercept survey at Philadelphia corner stores,16 and the results raise questions over tobacco industry claims about the importance of tobacco sales to small retailers.

Retailers’ perceptions that carrying tobacco products is important6 8 may stem from the tobacco industry which has a history of overstating the economic importance of tobacco.29 30 Tobacco manufacturers use economic arguments to persuade retailers around the globe to stock their products, even though retailers receive relatively little profit from these items.10 15 31 The tobacco industry encourages stakeholders to believe they are economically dependent on tobacco29 and should oppose retail regulation. Our study suggests removing tobacco from small retailers would be unlikely, in the majority of cases, to render these businesses unviable. However, variation in the prevalence of tobacco purchases across stores in our sample means some stores would be more affected than others. For instance, data from six retailers showed that tobacco made up more than a quarter of all transactions. Stores located in higher footfall locations and those that stock a broader non-tobacco product range, appear less likely to be affected by tobacco retail reduction policies, as tobacco sales make up a much smaller proportion of turnover in these stores.

Retailers’ belief that they need to sell tobacco for economic reasons are based on the premise that if they stopped selling tobacco, similar nearby businesses would continue to sell it, meaning they could lose sales to a neighbouring store.8 Yet policies to reduce tobacco outlet density could be applied equally to existing retailers, such as requiring all existing tobacco retailers to transition out of selling tobacco by a certain date and so would not disadvantage any store type.32

Our study has limitations. First, the expenditure data given by customers may not have been wholly accurate. However, any errors would have been non-differential and would not have biased estimates according to whether tobacco was purchased or not. Second, we did not include children under 12 years of age as participants and therefore may have overestimated the proportion of transactions that included tobacco. The generalisability of our results to other NZ districts and internationally is unknown, hence further studies with more geographically diverse samples are required. Future NZ studies could extend the data collection period to obtain larger sample sizes and hence more precise estimates. Lastly, we were unable to examine retailers’ profit from tobacco and non-tobacco items, or their actual turnover, thus we are unable to assess how their revenue might be affected by a tobacco retail reduction policy.

At most convenience stores in Dunedin, tobacco constitutes only a small proportion of items bought and tends not to be purchased along with other non-tobacco items. This study does not support arguments promulgated by tobacco companies that most small retailers rely on tobacco sales.