Statistics from Altmetric.com
More than half the world's countries have banned some forms of tobacco marketing,1 so tobacco pricing has become an ever more important component of both tobacco marketing2 and tobacco control efforts.1 Tobacco control policies based on price are a response to the price elasticity of smoking, with a large body of evidence showing that increased cigarette prices lead to reduced cigarette consumption.3 The effect of price on smoking appears to be even stronger in particular groups: one expert review of the effect of tax and price policies on tobacco use found strong evidence of greater responsiveness to price and tax increases among lower-income groups and ‘sufficient’ evidence of greater price elasticity by young people.4
The tobacco industry is characterised by a relative lack of price competition, where profits from higher prices can be invested in other marketing efforts, but if prices are set higher in order to maximise short-run profits, the resultant reduction in youth smoking would lead to a significantly lower number of smokers in the long run, lowering future profits.5 Choosing an optimal price is therefore important for tobacco companies interested in maximising long-term revenue. A large component of tobacco prices is tax, which has been said to be the ‘single most effective intervention to reduce demand for tobacco’.6 Tobacco companies have therefore often responded to tax increases by discounting prices,3 and evidence from the UK suggests that tobacco companies have kept the prices of discount brands lower by differentially shifting tax increases between brand segments.7 Industry documents also reveal extensive research into the price elasticity of smoking and discussion of varying approaches to reach smokers with different characteristics, such as those who are brand loyal and those who are price sensitive.5
A standard marketing approach under such conditions would be to use a segmentation strategy,8 possibly including lower prices for segments that are likely to be most price sensitive. Without restrictions on retail pricing, tobacco manufacturers and retailers could use such discount offers to grow target markets and profits1 and logically might attempt to target discounts to price-sensitive segments of the market. However, to our knowledge, there has been only limited investigation of whether tobacco prices are lower in areas where there are likely to be a higher percentage of price-sensitive smokers, despite growing evidence of such practices in the UK, the USA and Australia.9–13 Presumably in part due to the difficulty of collecting large-scale price data, previous studies of variations in price have usually been limited, either to reports of single promotions,9 limited types of outlets,12 ,13 one13 or two12 metropolitan areas, mid-size cities11 and areas around schools.10 Such limited analyses run the risk that variations in price may be explained by other factors; for example, one sample of prices in areas with medium-sized shopping centres found that differences in price according to the socioeconomic status of the area were no longer significant after controlling for the type of tobacco outlet.14
If targeted tobacco pricing is being used by tobacco manufacturers or retailers, Australia is a likely location for implementation of such a strategy: the country's wide-ranging restrictions on tobacco promotion have led to it being described as ‘one of the darkest markets in the world’ by a large tobacco manufacturer.15 Australian bans on tobacco advertising are broadly defined: the legislation does not permit any writing, picture, sign, symbol or audible message that promotes smoking or the purchase or use of tobacco products.16 As a result, we believe that multi-pack discounts, along with variation in pack pricing between brands and outlets, are the only forms of price promotion permitted in the Australian market. There is also some evidence that tobacco marketing strategies in Australia are used to inform strategies in other regions. The marketing director of British American Tobacco Australia has said that:
One of the best things we can offer the world is what we do best, which is how to work, maximize, proactively drive our market position in a market that's completely dark…We need to export that…we know we have a lot of expatriates who come down to Australia for learning…they can come here and learn these techniques and take them back to Europe or Latin America or to the United States or to Africa…17
If Australia is, as suggested in the quote above, a model for tobacco company learning, any evidence of targeted pricing strategies in Australia would be important for tobacco control in other countries, because it would suggest that any perceived success with pricing strategies in Australia would result in the extension of such strategies to other markets. The present study was therefore designed to investigate the extent of price discounting of cigarettes in Australia and to test for any variation in Australian tobacco prices across different geographic regions.
The study was conducted in New South Wales (NSW), the largest state in Australia, which comprises 32.1% of the Australian population.18 NSW legislation requires that retailers selling tobacco notify the state Ministry of Health (MoH), and as part of a Cancer Council NSW audit of tobacco retailers on the MoH register (henceforth ‘listed’ retailers), the price of a standard cigarette brand across NSW was collected. Since leading brands are likely to be carried by the largest number of outlets (including small outlets),19 the price of the market share leader (Winfield20) was chosen for analysis. (In a classification of Australian cigarette brands as ‘value’ (ie, lowest price), ‘mainstream (ie, mid-price) and ‘premium’ (ie, highest price), Winfield has been rated as a ‘mainstream’ brand.21) After several rounds of testing, an audit form was used to record, for each outlet, the price for both a twin pack (if sold) and a single pack of Winfield cigarettes. The audit also recorded the outlet type, since previous research has shown lower average prices at Australian tobacconists and supermarkets,22 compliance with tobacco display legislation (ie, no display of cigarettes or smoking products, only one price board, price listing of a maximum size, warning signs and for tobacconists, distance of tobacco display from public facing openings). The location of each audited outlet was coded for remoteness (since price competition in major cities was expected to result in lower prices in those areas); socioeconomic status (SES) of the postcode (reflecting evidence of greater price sensitivity in this group4), percentage under 18 in the postcode (reflecting evidence of greater price sensitivity in younger people4) and percentage in the postcode born in Australia (since smoking in Australia is higher among some migrant groups23). The data collected thus allowed assessment of whether retailers who fail to comply with tobacco notification and display legislation are offering lower cigarette prices.
At the time of the study, Cancer Council NSW had 10 regional offices and eight regions took part in the study. These regions covered 468 NSW postcodes containing 10 794 retailer records or 77.4% of the postcodes and 80.3% of the listed retailers. Postcodes for audit were randomly selected within each region until predefined sampling targets were reached for both the percentage of retailers and the percentage of postcodes (to ensure a sample that was broadly representative of each region). The percentage of retailer records within each region was calculated on the basis that we intended to audit 100% of the listed retailers within the sampled postcodes. For the two regions with the largest numbers of retailers (both located within Sydney, the largest city within the state), the sampling targets were 15% of listed tobacco retailers and 15% of postcodes within the region, and for regions with fewer retailers, 20% of listed tobacco retailers and 20% of postcodes. This process resulted in a total target sample of 100 postcodes. Five of the target postcodes were small rural localities in which no data collectors could be recruited, leaving a target sample of 2279 retailers in 95 postcodes—51 in regional areas and 44 in the Sydney metropolitan area. The target sample therefore encompassed 17.0% of the retailers and 15.7% of the 605 postcodes on the MoH list.
Data were collected between November 2012 and February 2013 (although only one outlet was audited in February) by 166 staff and volunteers associated with Cancer Council NSW. All data collectors received training, either face to face or by telephone. Postcodes with large numbers of tobacco retailers were shared between data collectors, with each asked to audit approximately 20 retailers, so that all listed tobacco retailers within the randomly selected postcodes were assigned for audit. Data collectors were also asked to look for and audit any tobacco retailers in their assigned area that did not appear on their summary list. Since auditors were expected to be more successful in identifying unlisted retailers in an area they knew, if they were not auditing their own postcode, auditors were also to look for, and audit, unlisted retailers in their own postcode. To assist the process of identifying unlisted retailers, auditors were asked to (1) conduct a systematic walk around known areas with multiple shops (ie, retail strips or shopping centres) looking for unlisted retailers, (2) use their knowledge (if any) of the area to recall and check retail outlets which might sell cigarettes (eg, convenience stores, bars, petrol stations and supermarkets) and (3) while auditing listed retailers in their assigned area, to look for any potential unlisted retailers. Auditors identified 664 outlets as inactive (not selling tobacco or no longer at the listed address) or as duplicate listings of outlets. The largest source of multiple listings was vending machines, since state legislation requires notification by both the vending machine operator and the licensee (since vending machines are only located in licensed premises). While correctly reflecting the legislation, such listings, if unaccounted for, inflate the actual number of tobacco outlets. A second source of multiple listings was records of two or more business names at the same address, presumably due to turnover of the business. All cases of multiple listings were reviewed and redundant records were excluded from analysis. After removing outlets no longer selling and duplicate listings, 1615 outlets remained, of which 1565 (96.9%) in 95 postcodes were audited.
Data collectors identified 174 unlisted retailers—120 in 52 of the 95 randomly selected postcodes audited and an additional 54 in 27 auditors’ own non-randomly selected postcodes. These 174 outlets were verified as unlisted by checking an updated list obtained from the MoH and added to the data for analysis, resulting in a total sample of 1739 retailers from 122 postcodes. Price data were collected for 1579 of the 1739 audited outlets. Auditor comments revealed a range of reasons for inability to provide price data, including vending machines being broken, the brand not available, only cigars being sold, refusal of staff to provide prices to someone doing a survey and someone not feeling confident to ask an apparently intimidating shopkeeper.
The dependent variable for the primary analysis was the outlet's minimum pack price for the reference brand (ie, half the twin pack price, if sold, or otherwise the single pack price). Audit results and auditor comments were reviewed to identify any breaches of tobacco display legislation. Independent variables were then created reflecting whether a retailer was found to have one or more violations of the point-of-sale display ban or other in-store provisions of the NSW Public Health (Tobacco) Act 2008 (‘Legislation breach’ in table 1), and also reflecting whether it was listed (as required) with the MoH (‘Listing with MoH’). The outlet type was classified using the classification of the outlet contained on the database, or for unlisted outlets, the outlet classification as judged by the volunteer auditor. The percentage in the postcode born in Australia and under 18 years were obtained from the Australian Bureau of Statistics (ABS) 2011 census data.24 The SES of the postcode was calculated using the ABS Socio-Economic Indexes for Areas.25 Remoteness of the postcode was calculated using the ABS Postcode to Remoteness Areas correspondence.26
Multiple linear regression was used to examine the associations between mean pack price and retailer characteristics with generalised estimating equation adjustment for the clustering of retailers within postcodes. Tests for linear trends were performed by substituting the nominal versions of independent variables with continuous or ordinal versions where appropriate. Robust variance estimators and an exchangeable working correlation matrix were used, with sensitivity analyses performed using various working correlation matrices. (Those results are omitted because differences were negligible.) The data were analysed using Stata 11.0 software.
Multi-packs were sold in 27.8% of audited outlets, with the average pack in a twin pack $1.32 (or 7.3%) cheaper than the average single pack price (p<0.001). The association between each independent variable and price, after controlling for other predictors, is shown in table 1. Prices varied significantly between different types of outlets (p <0.001), with tobacconists and supermarkets the cheapest, with mean pack prices of $16.09 and $16.51, respectively, compared with the overall mean price of $18.01. After controlling for other factors, there was a significant linear trend for lower prices in postcodes with a higher proportion of people aged under 18 years (p-trend<0.001). On average, prices were $0.62 lower (95% CI 0.32 to 0.92) per pack in postcodes with 25% or more of people aged under 18 years compared with postcodes with less than 16% aged under 18 years.
Prices also varied according to the level of disadvantage in the postcode, with the cheapest mean prices ($17.40) in the most disadvantaged postcodes. The relationship remained significant, but was not linear, after adjusting for other factors (p=0.027). Prices were also lower in postcodes with a higher proportion of the population born outside Australia, although the difference was only marginally significant (p=0.055). After controlling for other factors, and compared with postcodes with 90% or more born in Australia, on average, a pack was $0.32 cheaper (95% CI −0.06 to 0.69) in postcodes with fewer than 60% of residents born in Australia. Unexpectedly, prices in regional and remote areas were not significantly different from those in major cities; in fact prices were slightly (but not significantly) lower in regional and remote areas, even after taking into account the type of outlet and other factors. This was despite there being significantly more tobacco retailers per postcode—on average, more than three times more—in major city postcodes compared with regional and remote postcodes (p=0.003, table 2). There was no significant difference in price between listed and unlisted retailers or between retailers that breached the point-of-sale display ban and other in-store provisions of the Public Health (Tobacco) Act 2008 and those that complied. The intracluster correlation coefficient for pack price was 0.10.
Sensitivity analyses were conducted (1) restricting the dependent variable to the price of a single pack (rather than using the lowest pack price available at the outlet, including a pack price based on a twin pack purchase) and (2) excluding the 46 unlisted outlets with price information found in auditors’ own non-randomly selected postcodes. These analyses did not result in a material change in effect estimates. An additional sensitivity analysis, including month of audit, showed that it was not associated with price (p=0.5), nor was it a confounder of other adjusted price differences.
The results show that substantial price discounts are offered through the sale of twin packs in Australia. Evidence from Western countries suggests that the price sensitivity of demand for cigarettes is around −0.4.6 That is, for every 10% fall in the price of cigarettes, consumption can be expected to increase by about 4%. There is also evidence that tobacco purchase sizes are larger at outlets with lower prices,27 and larger purchases are likely to encourage increased consumption of cigarettes.2 Thus, the average 7.3% discount offered to a consumer by the purchase of a twin pack is likely to result in a meaningful increase in purchase and consumption of tobacco.
The existence of multi-pack and other discounts has been reported before,5 ,10 but these results also suggest that prices are systematically lower in locations with higher percentages of people under 18 and in lower socioeconomic areas—two groups that have been found to be more price sensitive.4 Contrary to expectations, we found that prices were not higher outside major cities (and in fact were lower, although not significantly so). This finding is in contrast with evidence that both petrol28 and groceries29 are typically more expensive in regional and remote areas in NSW and is consistent with tobacco companies offering targeted discounts to non-urban retailers to support lower prices, thereby encouraging smoking by price-sensitive consumers in those areas.
The results do not establish that lower prices in particular geographic areas are the result of tobacco industry activity, rather than the result of market forces in those areas. However, tobacco companies have discussed targeted promotions to tobacco retailers,2 and the finding of lower prices in regional and remote areas is surprising given the lower competition which would be expected to result from the lower number of tobacco retailers in such areas. Lower prices outside major cities are also inconsistent with the higher prices of food and petrol in such areas,28 ,29 as discussed above. Previous reports that cigarette prices are cheaper in lower socioeconomic areas9–13 could potentially be explained by lower retailer costs in those areas (eg, due to lower rents, thus allowing lower prices for a range of products). An exception is a study by Henriksen et al10 which found that the price of menthol cigarettes was negatively associated with the proportion of black students in nearby schools, but the price of the leading non-menthol brand was unrelated to any school or neighbourhood characteristics. Although we did not benchmark the cost of cigarettes against other consumer items in the same stores, comparing our finding of lower prices in non-metropolitan areas with Australian data on food and petrol prices adds to the Henriksen et al data, providing additional evidence that such price variations are not due to usual market mechanisms. Comparing the price of cigarettes with consumer staples such as food and petrol may provide a useful way for future research to assess whether tobacco prices are consistent with variation in other consumer products, and thus reflect local retail conditions (eg, retail rental prices and/or consumer willingness to pay), or instead are consistent with targeted price discounts. Whether they are the result of industry activity or some other factor, the price differences identified in this study, which offer lower prices to price-sensitive smokers, have the potential to undermine attempts to lower smoking rates. As most governments have targets to reduce smoking rates, these findings raise the issue of whether there should be public policy interventions to counter discounting of cigarettes (as implemented in Providence, the capital of Rhode Island, in 201330) and disclosure of any promotional discounts offered to retailers.
There are some limitations on the generalisability of these results. The results are limited to one brand (albeit the best-selling brand in Australia). While the sample was large, and the randomly selected postcodes were chosen from regions containing more than 80% of NSW retailers, we did not audit most of western NSW, and so the findings cannot be applied with confidence to that area (which is characterised by sparsely populated areas and small towns). However, the sample included both remote and rural areas, and beyond the expected finding of fewer tobacco retailers in those areas, we did not find large differences between urban and non-urban areas, so we have no reason to expect that the findings would be different if we had sampled all regions. In addition, the price model allowed for remoteness of the postcode, so unless outlets in Western NSW are systematically different from the remote outlets included in the model, these results should be able to be extrapolated to remote areas of NSW. An additional limitation of our study is that the process used to identify unlisted retailers was unlikely to identify all unlisted retailers, so our sample of unlisted retailers represents an unknown percentage of the actual percentage of unlisted retailers. However, listed retailers were a relatively small percentage of the total sample (10.1%, as shown in table 1) and were not associated with lower prices, so our inability to detect all unlisted retailers is unlikely to influence the results in any meaningful way.
It should also be noted that the audit (from November 2012 to February 2013) was mostly conducted after the implementation of plain packaging on December 1. While plain packaging of tobacco in Australia may limit the generalisability of the results, it is possible that the effects of any targeted price discounts would be greater in markets without plain packaging.
Policies banning point-of-sale displays of tobacco and requiring plain packaging of cigarettes have limited the ability of tobacco companies to promote their products in Australia. However, this study shows that substantial multi-pack discounts are widely available in the Australian market, and discount deals are likely to increase impulse purchases and encourage smokers to consume more than they might otherwise.2 The evidence of systematic variation in prices across NSW is consistent with selective discounts being offered to maintain smoking among consumer segments, which are likely to be more price sensitive. The guidelines of the Framework Convention on Tobacco Control (FCTC) recommend banning all forms of price promotion31 and identify that discounts and other retail merchandising activities are forms of promotional activity which are covered by the FCTC.2 However, these results show that multi-pack and regional discounts persist in the NSW market and are consistent with discounts being targeted to price-sensitive smokers. Since industry documents suggest that tobacco marketing strategies in Australia inform practice in other countries,15 such a finding is a concern for tobacco control in other markets.
Several authors have argued for the use of floor prices (ie, setting a minimum retail cost for tobacco sales) to limit discounting of tobacco.32 ,33 However, a floor price policy has the potential to benefit the tobacco industry or retailers through higher profits.34 One possible option to counter selective discounts is therefore a price-cap system, which would limit the price that manufacturers can charge to reduce the potential for price differentiation while assuring through appropriate excises that prices do not fall below a certain level.35 Another possible policy is to compel manufacturers to disclose their promotional activities, as required in Canada,35 and thereby ensure that any price discounts to particular retailers or regions are publicly disclosed, and thus open to scrutiny and debate.
These results suggest that tobacco companies and/or retailers may be selectively using price discounts to keep smoking rates as high as possible. Even if these strategies are currently restricted to Australia, it is likely that any strategy the tobacco industry views as successful will be quickly replicated in other markets. It may be time for governments to develop policies that counter price discounts targeted at price-sensitive consumers in order to decrease smoking and achieve corresponding increases in consumers’ quality of life.
What this article adds
Provides evidence of systematic variations in the price of cigarettes that are not explained by normal market forces.
Finds that cigarette prices are lower in areas where there are likely to be more price-sensitive smokers, which will counter government efforts to decrease smoking by tax increases.
While the results cannot establish that price variations are the result of targeted discounts offered by tobacco companies, they are consistent with such strategies, and thus support mandatory disclosure of tobacco company promotional efforts.
The results contribute to the debate on the need for policy interventions on prices for tobacco as a means to counter smoking.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.