Tobacco Purchasing in Australia during Regular Tax Increases: Findings from the International Tobacco Control Policy Evaluation Project

Abstract

Objective

We examined Australian tobacco purchasing trends, the average self-reported price paid within each purchase type, and the association between type of tobacco product purchased and participant characteristics, including quit intentions, between 2007 and 2020.

Methods

We analysed data collected from adults who smoked factory-made and/or roll-your-own cigarettes (RYO) in nine waves (2007–2020) of the International Tobacco Control Policy (ITC) Evaluation Project Australia Survey (N_sample=5,452, N_observations=11,534). The main outcome measures were type of tobacco products purchased: RYO, carton, pack or pouch size and brand segment. Logistic regression, fit using generalized estimating equations, was estimated the association between the outcome and participant characteristics.

Results

The reported price-minimising purchasing patterns increased from 2007 to 2020: any RYO (23.8% to 43.9%), large-sized pack (2007: 24.0% to 2016: 34.3%); shifting from large to small-sized packs (2020: 37.8%), and economy brand (2007: 37.2% to 2020: 59.3%); shifting from large (2007: 55.8%) to small economy packs (2014: 15.3% to 2020: 48.1%). Individuals with a lower income, a higher nicotine dependence level, and no quit intention were more likely to purchase RYO and large-sized packs.

Conclusion

RYO, large-sized packs, and products with a low upfront cost (e.g., small RYO pouches and small-sized economy brand packs) may appeal to people on low incomes. Australia’s diverse tobacco pack and pouch sizes allow the tobacco industry to influence tobacco purchases. Standardizing pack and pouch sizes may reduce some price-related marketing and especially benefit people who have a low income, are highly addicted and have no quit intention.

BACKGROUND

Australian cigarette prices are among the highest in the world(1) due to tobacco excise increases (25% in 2010 and 12.5% annually from 2013 to 2020), and biannual inflation adjustment.(2,3) An interrupted time-series study reported an immediate decline in smoking prevalence after the 2010 tax increase among lower socioeconomic status (SES) groups. The impact weakened over time, likely due to a shift to less expensive roll-your-own (RYO) tobacco.(4,5)

The tobacco industry uses various strategies to diminish the impact of tobacco taxes,(6–8) such as introducing new brand variants(9) and relaunching super-value brands.(10) For example, the proportion of Australian economy brands doubled from 20% in 2013 to 40% in 2015.(11) The three companies that supplied 95% of the Australian market in 2020 (British American Tobacco Australia Limited, Imperial Tobacco Australia Limited, and Philip Morris Limited)(12) expanded the ‘standard’ factory-made cigarette (FMC) pack sizes (20, 25, 30, 35, 40 and 50 cigarettes) to include packs containing 21, 22, 23, 26, 32 and 43 cigarettes.(2,8) The RYO pouch size range also expanded from 30g, 35g, 40g, and 50g to include 15g, 20g, 25g, 27g, 45g, 55g. Multi-pack bundles (twin packs and cartons) are also used.(8,10) Since the 25% tobacco tax increase, these strategies increased opportunities to reduce the upfront cost of purchasing tobacco,(13) to purchase at lower unit prices, give the appearance of ‘bonus’ cigarettes, and make calculating unit pricing difficult.

Higher prices increase quit motivation,(1,14,15) and reduce cigarette sales,(16) but some price-sensitive consumers use cost-minimising strategies to reduce the impact of tobacco tax increases, such as reducing the quantity smoked, switching to cheaper tobacco products, and using discounts (bulk purchasing).(17) An analysis of the International Tobacco Control Policy Evaluation Survey (ITC Project) in the United Kingdom, Canada, the United States, and Australia found that lower SES consumers were 25% more likely to use multiple tax avoidance behaviours compared to higher SES consumers,(18) and these behaviours were associated with 28% reduction in smoking cessation regardless of SES status.(17) An Australian study found that lower income, less education or living in a lower SES neighbourhood was associated with using cost-minimising strategies after the 25% tobacco tax increase in 2010.(19) However, this research only included data from one state and did not examine the effects of the series of tax increases enacted after 2013. Since the lower cost of RYO and a wide range of cigarette pack and RYO pouch sizes may have reduced the effectiveness of tax increases, further research on tobacco purchasing and prices paid during a period of regular tobacco tax increases could assist understanding the combined impact of tobacco tax increases and industry strategies on consumer behaviour and quit intentions. Therefore, this study aimed to:

1. examine tobacco purchasing behaviour (purchasing in cartons, economy brand, small or large pack, or RYO) and the price between 2007 and 2020;

2. estimate the correlation between price within each purchase type and tobacco excise over the study period; and

3. estimate the relationship between participant characteristics and purchasing behaviour.

METHODS

Sample

We analysed data from nine waves (2007–2020) of the Australian arm of the ITC Project, a longitudinal cohort of adults who currently or recently smoked in Australia. The data were collected via computer-assisted telephone interviewing or online, with data collected exclusively online since 2016. Each wave, respondents lost-to-follow-up were replenished from the same sampling frame to maintain sample size. Attrition increased from around 30% in the first six waves to about 43% when phone-only respondents were dropped. Sampling weights for each participant were calibrated to align the sample to the Australian smoking population using official national figures. Initial cross-sectional weights were then rescaled to sum to the overall sample size.(20,21) Detailed information on the survey is available elsewhere.(22,23) The analytical sample for this study comprised 5,452 adults who completed at least one survey and smoked at least monthly. The average number of waves per participant was 2.2 (SD=1.6). These participants contributed 11,534 observations (N2007/2008=1,896; N2008/2009=1,375; N2010–2011=1,153; N2011–2012=1,104; N2013=1,093; N₂₀₁₄=1,090; N₂₀₁₆=1,339; N₂₀₁₈=1,253; N₂₀₂₀=1,231) during the study period. We created two additional datasets of participants who purchased: 1) FMCs to analyse last type of tobacco product purchased, pack sizes and brands (N_all=9,030; N_2007/2008=1,460; N_2008/2009=1,149; N_2010–2011=884; N_2011–2012=889; N₂₀₁₃=831; N₂₀₁₄=833; N₂₀₁₆=1,061; N₂₀₁₈=985; N₂₀₂₀=938); and 2) RYO to analyse pouch sizes (N_all=2,169; N_2007/2008=268; N_2008/2009=210; N_2010–2011=188; N_2011–2012=205; N₂₀₁₃=246; N₂₀₁₄=245; N₂₀₁₆=261; N₂₀₁₈=262; N₂₀₂₀=284). The data collections were approved by Human Research Ethics Committees of the University of Waterloo, Canada (REB#13978, REB#17469, and REB#20803/30878), Cancer Council Victoria (IRB HREC0211 and IRB HREC1063) and the University of Queensland (IRB #2016000330), who granted ethics exemption for this analysis (2022/HE001464).

Measures

Current tobacco product use and last purchase was categorized into: “FMC only”, “RYO only” and “both (mainly FMC, FMC and RYO similar, and mainly RYO)”.

FMC purchase types were: by pack (FMC-P), or carton of 5+ packs (FMC-C). Because we restricted our analysis to legal sales, we excluded single cigarette purchases (0.35%).(24) Packs were categorized into small-size (20s, 21s, and 22s), mid-size (23s, 25s, 26s, 30s, 32s, and 35s), and large-size (40s and 50s).

RYO pouches were categorized into small (15g, 20g, 25g and 27g), mid (30 and 35g, 40g and 45g), and large (50g and 55g). The smallest pouch size available was 30g until 2010, 25g from 2011 to 2015, 20g from 2016 to 2017, and 15g from 2018.(2)

One researcher (AC) classified FMCs into four brand market segments: super-value, value, mainstream, and premium based on RRPs listed in trade magazines,(25–33) and ordered brands by average price per stick in each wave; each market segment cut-point was determined from the listing of key brands by market segment in Tobacco in Australia(34) and examples of market segments from Euromonitor International reports (Supplemental Table 1-B).(9,35–41) Where variants within a brand family crossed price segments, we classified them into separate market segments rather than averaging them. Because the trade magazine did not provide RRPs in 2020, we used company price lists from wholesaler websites (BATA, PM, and ITA) or the price list created for 2018 or the market segment list for 2019 from Tobacco in Australia for the remaining brands (5% of market share). A senior researcher (MS), an expert on Australian tobacco brands, checked and confirmed the categorisation.

Potential cigarette tax avoidance/evasion was defined as sourcing cigarettes from duty-free stores, out of country, by mail, phone, or internet, or “someone else” (informal sellers).(42,43) Average cigarettes per day (CPD) was coded as 0 (1–10 cigarettes), 1 (11–20 cigarettes), 2 (21–30 cigarettes), and 3 (>30 cigarettes). Time to first cigarette (TTFC), determined by asking “How soon after waking do you usually have your first smoke?” was classified as 0 (>60 minutes), 1 (31–61 minutes), 2 (6–30 minutes), and 3 (5 or fewer minutes). The heaviness of smoking index (HSI), a measure of nicotine dependence,(44) was calculated as the sum of CPD and TTFC and grouped into low (0–2), moderate (3–4) and high (5–6).(45) We calculated the price per stick of last tobacco purchase from price paid and number of sticks or RYO weight bought. Previous research indicates most RYO cigarettes contain at least 0.5g tobacco,(46) so we calculated price per stick for both 0.7g and 0.5g. The calculations relied on multiple responses, which produced inconsistencies and missing data. Therefore, we exclude improbable responses from price analyses based on: 1) potential tax avoidance/evasion; 2) all prices per FMC stick or 0.8g of RYO below the excise value ($2012) each year (Supplemental Table 2)(47); 3) FMC packs containing <20 cigarettes or >50; 4) RYO pouches containing <20g or >55g. All prices were converted to 2012 AUD using Consumer Price Index (CPI) data.(48)

We calculated weight per RYO cigarette from a) RYO pouch weight purchased, b) usual number of days to smoke this amount, and c) average CPD. We excluded responses with a weight per cigarette <0.1g or >1.0g, and responses with missing data (N=196).

Sociodemographic measures were sex (male and female), age, household annual income (low: <$30,000 from 2007 to 2014; <$45,000 from 2016 to 2020, moderate: $30,000–59,999 from 2007 to 2014; $45,000–74,999 from 2016 to 2020, and high:$60,000+ from 2007 to 2014; $75,000+ from 2016 to 2020, and “don’t know”), and education level (low: high school or less, moderate: technical/grade/college completion/some university, and high: university degree). “Don’t know” was treated as missing (N=30).

Intention to quit, derived from “Are you planning to quit smoking?” was coded into: in the next month, in the next 6 months, after 6 months, no intention, and don’t know.

Analysis

We used STATA (SE V.17) to estimate weighted proportions of purchasing behaviours for the Australian smoking population using the svy command to account for the complex sampling design. We used the initial rescaled cross-sectional weight corresponding to each participant’s first wave of appearance to make the weights consistent within respondents with multiple measurements.(20,23) The crude association between the nominal prices and tobacco excise and customs duty (in 2012 AUD) was estimated using linear generalised estimating equations (GEE)(49) regression using a Gaussian distribution, identity link function and exchangeable correlation. Logistic regression fit using GEE with the xtgee command was used to explore factors associated with purchasing behaviours. GEE models were fit with a logit link, a binomial family, and an exchangeable correlation structure and were adjusted for wave, survey mode (web vs. telephone), age, sex, annual household income, education, smoking frequency, nicotine dependence, quit intention, and age-year interaction. We used survey wave as a co-variate rather than year to account for the unequal spacing between survey waves since describing any effects as simple linear increases (or decreases) over the study period may have masked effects (e.g., the 25% excise tax increase in 2010 vs. the annual 12.5% increase from 2013 to 2020). A multinomial regression analysis with the xtmlogit command was performed for pack size and brand purchased. The xtset command specified the respondents formed the random intercept for the xtmlogit model. The GEE and multinomial analysis were conducted in two ways – weighted to the Australians smoking population, and unweighted. When the year-by-age interaction effect for purchasing behaviour was statistically significant, model-based marginal probabilities and confidence intervals were estimated using the margins command. There was no substantial multicollinearity among the covariates across models as determined by variance inflation factors being less than 5.(50) We included variables significant at an alpha level of 0.05 in adjusted analyses.

RESULTS

Table 1 shows respondent characteristics at their first survey wave. Half (50.4%) were female, about one-third (30.2%) had low annual household income, and 46.3% had low education level. Most (91.7%) smoked daily, 69.1% smoked FMC exclusively, 11.6% smoked RYO exclusively, and 21.2% had no plan to quit smoking. Among those who purchased FMC, the majority (81%) purchased in packs (not cartons). Mean (standard deviation) age and HSI were 47.1 (14.0) and 2.6 (1.6), respectively.

Table 1.

Descriptive characteristics for participants when they first participated in the International Tobacco Control (ITC) study, 2007–2020

Variable	Categories	N	%
Age*			47.1(14.0)
Sex	Male	2 747	49.6
	Female	2 705	50.4
Annual household income	Low	1 541	31.3
	Moderate	1 369	27.9
	High	2 004	40.8
	Don’t know	186	3.7
Education level	Low	2 512	46.3
	Moderate	1 755	32.4
	High	1 155	21.3
Smoking frequency	Daily	5 000	91.7
	Weekly	329	6.0
	Monthly	123	2.3
Heaviness of smoking*			2.62(1.60)
	Low	2 124	42.3
	Moderate	2 322	46.2
	High	580	11.5
Tobacco product use	Exclusive FMC use	3 761	69.1
	Both FMC and RYO use	1 049	19.3
	Exclusive RYO use	630	11.6
Tobacco tax avoidance/evasion	No	5 207	98.4
	Yes	86	1.6
Tobacco product as the last purchase	FMC purchase	4 171	81.8
	RYO purchase	931	18.3
Tobacco purchase types†	Pack purchase	3 377	81.0
	Carton purchase	794	19.0
Quit intention	No intention	1 108	21.2
	In next month	705	13.5
	In next 6 months	1 317	25.1
	After 6 months	1 823	34.8
	Don’t know	287	5.5
Total		5 452	100

N, unweighted sample size; %, unweighted percentage; FMC, factory-made cigarettes; RYO, roll-your-own cigarettes

^*mean (standard deviation); low: mean=1.1 and SD=0.9; Moderate: mean=3.4, SD=0.5; High: mean=5.2, SD=0.4

^†among adults who aged 18+ and purchased FMC as their last purchase

Figure 1 shows percentages of survey respondents reporting purchasing each RYO pouch size and FMC pack size, within each brand market segment, and the average self-reported prices paid per stick over the study period. The detailed sample size and weighted proportions, and average nominal prices paid per stick are presented in Supplemental Table 3 and Supplemental Table 4. Figure 1-A shows the proportion of people purchasing large-sized RYO pouches (50g and 55g) decreased from 74.4% in 2007/2008 to 18.7% in 2020. The proportion reporting small-sized pouch purchases (25g and 27g) increased about fivefold from 10.4% (2013) to 54.7% (2020). Purchasing 20g pouches increased from 5.1% (2016) to 19.0% (2020). Self-reported average RYO price per 0.7g stick significantly increased from $0.35 (2007/2008) to $1.09 (2020, p for trend<0.001). Estimated average cigarette weight significantly decreased from 0.48g (2007/2008) to 0.38g (2020, p for trend<0.001). Figure 1-B shows the proportion of participants purchasing FMC-Ps that bought economy brands, including super-value and value brands, increased from 37.2% (2007/2008) to 59.3% (2020), with the proportion purchasing super-value brands increasing sharply from 10.8% (2014) to 24.3% (2016) before falling to 9.2% (2020). The proportion purchasing premium brands increased from 13.7% (2018) to 24.3% (2020).

Figure 1.

Purchasing behaviours and the average prices paid by respondents, 2007–2020. (A) The smallest pouch size available was 30g until 2010, 25g from 2011 to 2015, 20g from 2016 to 2017 and 15g from 2018. (B) Economy brand includes super-value and value brands. (C) Small-sized pack includes 20s (21s, 22s); middle-sized pack includes 25s (23s, 26s) and 30s (32s, 35s); large-sized pack includes 40s and 50s. % denotes weighted proportion. *Average prices paid by respondents for roll-your-own (RYO) products. ‡Average prices paid by respondents for factory-made cigarettes (FMCs) in each price segment; economy includes super-value and value brands. §Average prices per stick paid by respondents for FMCs in large versus medium versus small pack.

Figure 1-C shows the proportion of participants purchasing large-sized packs increased from 22.8% (2008/2009 to 32.2% (2013). From 2014 to 2020, the proportion purchasing small-sized packs increased from 15.8% to 37.7%, becoming the most purchased pack size in 2020. The average price per cigarette for large-sized packs was consistently lower than for small or mid-sized packs.

Supplemental Table 5 shows the correlations between price paid within each purchase type and the tobacco excise/duty. Price paid within each purchasing behaviour was positively correlated with level of tobacco excise/duty, and all coefficients (β) were >1, suggesting the increase in price paid for tobacco is more than the increase in tobacco excise/duty. However, the difference between purchasing behaviours was moderate.

Table 2 shows the results from GEE and multinomial logistic regression of purchasing behaviours and covariates. Unweighted results are shown in Supplemental Table 6. Compared to exclusive FMC use, any RYO use significantly increased since 2013 (aOR₂₀₂₀=3.10, 95% CI=2.24–4.30, vs. 2007/2008). Being male (aOR=1.76, 95% CI=1.492.08 vs. female), on a lower income (aOR_low=1.82, 95% CI=1.50–2.21; aOR_moderate=1.41, 95% CI=1.19–1.68 vs. high), and a high HSI score (aOR_high=1.27, 95% CI=1.06–1.53 vs. low) was associated with RYO use. No intention to quit smoking (aOR_{in next month}=0.66, 95% CI=0.54–0.80; aORin next 6 months=0.73, 95% CI=0.63–0.84, aORafter next 6 months=0.83, 95% CI=0.74–0.93, vs. no intention) was associated with RYO use. There were significant age-year interactions for those using RYO (vs. exclusive FMC use, p=0.017). The marginal estimates of RYO use at 30-years-old significantly increased from 23.8% (2007/2008) to 55.2% (2020), whereas the estimates for a 60-year-old started around same percentage (21.0%) and ended at 36.4% (Supplemental Table 7).

Table 2.

The association between participant characteristics and purchasing behaviours among those who purchased any FMC, 2007–2020

		Any RYO use (N=3,341) vs. ref: exclusive FMC use (N=8,139)		Carton purchase (N=1,850) vs. ref: pack purchase (N=7,035)		Large-sized pack* (N=1,807) vs. ref: mid-sized pack (N=3,725)		Small-sized pack* (N=1,398) vs. ref: mid-sized pack (N=3,725)		Economy pack† (N=3,938) vs. ref: premium (N=1,539)		Mainstream pack (n=2,928) vs. ref: premium (N=1,539)
		aOR	95% CI	aOR	95% CI	aOR	95% CI	aOR	95% CI	aOR	95% CI	aOR	95% CI
Wave (ref: 2007/2008)	2008/2009	0.96	0.87–1.05	0.98	0.84–1.13	0.84	0.57–1.23	1.27	0.82–1.97	1.68	1.16–2.42	1.27	0.86–1.89
	2010/2011	1.03	0.90–1.18	0.83	0.66–1.03	1.03	0.63–1.71	1.01	0.61–1.68	0.72	0.41–1.26	2.43	1.42–4.18
	2011/2012	1.08	0.91–1.28	0.86	0.69–1.07	1.55	0.88–2.72	2.42	1.42–4.15	1.05	0.55–2.00	2.31	1.09–4.91
	2013	1.47	1.21–1.80	0.81	0.62–1.07	2.62	1.42–4.83	1.79	0.99–3.20	1.77	0.90–3.47	2.26	1.10–4.64
	2014	1.46	1.18–1.80	0.80	0.60–1.08	2.62	1.32–5.22	1.48	0.77–2.85	4.36	2.05–9.24	1.74	0.78–3.88
	2016	2.44	1.91–3.11	0.58	0.41–0.81	4.65	2.14–10.09	2.85	1.42–5.76	8.71	3.45–21.97	1.60	0.61–4.19
	2018	2.67	1.98–3.60	0.46	0.32–0.66	2.84	1.16–6.95	5.65	2.52–12.67	8.68	3.18–23.7	3.39	1.12–10.27
	2020	3.10	2.24–4.30	0.47	0.30–0.72	3.27	1.02–10.44	13.41	5.32–33.78	3.50	0.87–14.09	0.52	0.13–2.14
Age		0.98	0.98–0.99	1.03	1.02–1.04	1.03	1.01–1.05	0.99	0.98–1.01	1.01	0.99–1.03	0.98	0.96–0.99
Gender (ref: female)	Male	1.76	1.49–2.08	0.96	0.78–1.18	0.27	0.17–0.43	1.42	0.98–2.06	0.35	0.22–0.58	0.98	0.58–1.68
Annual income level (ref: high)	Low	1.82	1.50–2.21	0.65	0.52–0.82	2.39	1.43–3.99	1.82	1.12–2.97	5.24	2.93–9.38	1.13	0.48–2.64
	Moderate	1.41	1.19–1.68	0.69	0.52–0.92	2.09	1.31–3.34	1.56	1.00–2.44	2.83	1.65–4.86	1.11	0.61–2.01
	Don’t know	2.06	1.43–2.98	0.51	0.31–0.84	3.21	0.99–10.37	2.19	0.83–5.78	2.38	0.90–6.30	0.79	0.26–2.37
Education level (ref: high)	Low	0.97	0.76–1.23	1.11	0.83–1.48	5.22	2.80–9.70	0.85	0.51–1.41	15.1	7.29–31.32	16.95	8.05–35.7
	Moderate	1.00	0.77–1.30	0.97	0.73–1.28	2.15	1.20–3.88	0.72	0.43–1.22	5.17	2.37–11.27	5.85	2.86–11.9
Smoking frequency (ref: non-daily)	Daily	1.01	0.74–1.37	2.27	1.46–3.52	8.13	3.25–20.34	0.35	0.19–0.63	5.6	2.54–12.36	2.82	1.43–5.57
Nicotine dependence (ref: low)	Moderate	1.06	0.95–1.19	1.77	1.50–2.07	2.27	1.54–3.36	0.47	0.34–0.67	2.43	1.62–3.65	1.17	0.71–1.92
	High	1.27	1.06–1.53	2.80	2.19–3.56	8.36	4.20–16.65	0.15	0.06–0.33	5.99	2.75–13.02	1.99	0.84–4.74
Quit intention (ref: no intention‡)	In next month	0.66	0.54–0.80	0.47	0.37–0.60	0.43	0.23–0.80	2.88	1.70–4.89	0.85	0.46–1.57	1.70	0.91–3.18
	In next 6 months	0.73	0.63–0.84	0.61	0.49–0.74	0.91	0.57–1.43	1.48	0.93–2.37	1.30	0.76–2.21	1.76	1.01–3.04
	After 6 months	0.83	0.74–0.93	0.83	0.70–0.99	1.05	0.71–1.55	0.96	0.61–1.51	0.90	0.53–1.52	1.64	0.99–2.69
	Don’t know	0.89	0.72–1.09	0.94	0.73–1.22	0.95	0.42–2.11	0.88	0.38–2.02	1.33	0.52–3.44	1.24	0.49–3.15

Survey mode and variables in the list were adjusted for in all models. RYO, roll-your-own cigarette; FMC, factory-made cigarette; N, unweighted sample size (observation); aOR, adjusted odds ratio; CI, confidence interval; Only purchased from AU store-based sources.

^*Middle-sized pack includes 25s (23s,26s) and 30s (32s,35s); Large-sized pack includes 40s and 50s; Small-sized pack includes 20s (21s, 22s);

^†Economy brand includes super-value and value brands;

^‡no intention includes not planning to quit

Compared to purchasing in packs, the odds of purchasing cartons decreased over the study period, but only significantly so in 2016, 2018, and 2020 (aOR₂₀₂₀=0.58, 95% CI=0.41–0.81 vs. 2007/2008). Those who had a high household income (aOR_low=0.65, 95% CI=0.52–0.82 vs. high), smoked daily (aOR=2.27, 95% CI=1.46–3.52 vs. non-daily), and had a high HSI score (aOR_high=2.80, 95% CI=2.19–3.56 vs. low) were more likely to purchase cartons than packs. No intention to quit was associated with purchasing cartons (aOR_{in next month}=0.47, 95% CI=0.37–0.60 vs. no intention).

Compared to purchasing middle-sized packs, the odds of large-sized and small-sized pack purchases increased from 2011/2012 onwards (aOR_{large-sized in 2020}=3.27, 95% CI=1.02–10.44; aOR_{small-sized in 2020}=13.41, 95% CI=5.32–33.78, vs. 2007/2008). Older age (aOR=1.03, 95% CI=1.01–1.05), female (aOR_male=0.27, 95% CI=0.17–0.43 vs. female), lower household income (aOR_low=2.39, 95% CI=1.43–3.99; aOR_moderate=2.09, 95% CI=1.31–3.34 vs. high), lower education level (aOR_low=5.22, 95% CI=2.80–9.70; aOR_moderate=2.15, 95% CI=1.20–3.88 vs. high), daily smoking (aOR_daily=8.13, 95% CI=3.25–20.34 vs. non-daily) and high HSI score (aOR_high=2.80, 95% CI=2.19–3.56 vs. low) were associated with purchasing large-sized packs, compared to middle-sized packs. Lower household income (aOR_low=1.82, 95% CI=1.12–2.97; aOR_mederate=1.56, 95% CI=1.00–2.44 vs. high), occasional smoking (aOR_daily=0.35, 95% CI=0.19–0.63 vs. non-daily), lower nicotine addiction (aOR_high=0.15, 95% CI=0.06–0.33 vs. low), and intending to quit smoking in the next month (aOR=2.88, 95% CI=1.70–4.89 vs. no intention) were associated with purchasing small-sized packs, compared to middle-sized packs.

Compared to premium brands, economy brand purchases significantly increased in 2008/2009 and from 2014 to 2018 (aOR₂₀₁₈=8.68, 95% CI=3.18–23.7 vs. 2007/2008). Being female (aOR_male=0.35 95% CI=0.22–0.58 vs. female), on a lower income (aOR_low=5.24, 95% CI=2.93–9.38; aOR_moderate=2.83, 95% CI=1.65–4.86 vs. high), lower education (aOR_low=15.1, 95% CI=7.29–31.32; aOR_moderate=5.17, 95% CI=2.37–11.27 vs. high), and high HSI score (aOR_high=5.99, 95% CI=2.75–13.02, vs. low) was associated with purchasing economy brands, compared to premium brands. There were significant age-year interactions for purchasing economy brands (vs. premium brand purchased, p<0.001). The marginal estimates of economy brand purchase for a 30-year-old significantly increased from 32.0% (2007/2008) to 55.4% (2020), while estimates for a 60-year-old remained at similar percentages (45.2% in 2007/2008 to 52.7% in 2020 over the study period, Supplemental Table 7). There were no year-by-education and year-by-income interaction effects for any outcome (data not shown).

DISCUSSION

Tobacco purchasing behaviours among Australian adults notably shifted from exclusive FMC purchasing to exclusive RYO or both RYO and FMC. We observed a shift from large-sized RYO pouches (50g and 55g) to small-sized pouches (25g and 27g), reflecting introduction of smaller pouch sizes in 2013.(2) These purchasing changes reflect the impact of industry marketing strategies on consumer behaviour, including rapid introduction of new RYO brand variants in smaller pouch sizes (and lower upfront purchase price) compared to the price of small FMC-P. (4) For example, the RRPs of the cheapest available product in small pouch size over the study period (Winfield 30g in 2007, Holiday 30g in 2010, Orlando 25g in 2013 and JPS 25g in 2016) were under $A20. An Australian discrete choice experiment study found that $20 was a critical price point at that time associated with plans to quit smoking.(51) BATA and ITA (90.8% of the Australian RYO market in 2020(48)) progressively introduced popular FMC brands as RYO, providing brand-loyal consumers with a low-cost alternative to FMC. Brands such as JPS (21.4%), Winfield (16.1%), and Rothmans (11.2%) led the RYO market in 2020, with traditional RYO brands decreasing from 80% of the RYO market in 2001 to 36% in 2016.(4)

We also found the gap in the reported average FMC-RYO tobacco prices paid per stick decreased from 24% in 2018 to 9% in 2020 (absolute change), demonstrating the positive impact of the RYO-FMC equalising tax measures implemented from 2017 to 2020. Nevertheless, the average RYO price per 0.7 g stick remained cheaper than the average FMC stick in 2020. Moreover, the weight of RYO cigarettes decreased from 0.48g (2007/2008) to 0.38g (2020, p for trend<0.001), consistent with previous ITC Project research covering 2006–2015 that found RYO cigarettes in the UK and Australia, where cigarette prices were relatively high, weighed less than in Canada and the USA.(46) That study also found the mean weight of each RYO cigarette in the UK and Australia decreased significantly by about 2% per year. Unfortunately, this reduction in tobacco use may not translate into measurable health benefits and could increase harm if it reduces motivation to quit by reducing cost.(52) Similar to the trend towards purchasing smaller RYO pouches, the percentage of large-sized FMC-P purchases increased from 2007 to 2016; shifting from large to small-sized FMC-P purchases in 2020, with participants on low incomes more likely to report buying them. Our finding of an association between purchasing small packs and lower nicotine dependence and having quit intentions, is consistent with purchasing patterns observed in hypothetical purchasing experiments.(53) Small packs may be a way to regulate consumption.(54) However, both smaller packs with a lower upfront price and larger packs with a lower stick price may encourage price-sensitive consumers to continue smoking by reducing cost barriers to continued smoking.

Our observation of a growing proportion of purchasing in the RYO, FMC large-size pack, and economy brand products is consistent with other research showing a growing market share of cheaper tobacco products.(2) Purchasing FMCs in cartons is a cost-minimising strategy because of the lower unit price than individual packs, but was more likely for those with a high income or high nicotine dependence. The high upfront cost may prevent people on low incomes from purchasing cartons. Our study revealed that consumers on low incomes and with high nicotine dependence are more likely to purchase RYO, large-sized FMC packs, and economy brands. Any RYO use increased 137% from 2007/2008 to 2020 among young consumers, while there was a more moderate increase among older consumers, consistent with other research that found exclusive RYO use was lowest among 18–29 year-olds and highest among those who aged 60+ years in 2004 and 2007,(4) with the pattern reversing in 2010, 2013 and 2016. Purchasing large-sized packs and economy brands also increased (47% and 57%, respectively) between 2007/2008 and 2020 among aged young consumers, while these purchases remained steady among older consumers. Our results suggest that RYO, large-sized packs, and economy brand products became less costly substitutes for standard-sized FMCs among young adults. These cost-minimising strategies may be reducing the effectiveness of Australia’s high tobacco taxes as motivation to quit smoking. Economy brands made up an increasing proportion of FMC purchases, with super-value brands increasing sharply from 2014 to 2016 before falling in 2020. This pattern corresponds to the marketing strategies of launching new super-value brands, re-pricing cigarettes between market segments and offering brand variants at different prices. These strategies ameliorate negative perceptions of down-trading and facilitate movement to cheaper products as an alternative to quitting smoking.(10) For example, Rothmans (BATA), a top-selling brand, was a premium cigarette in Australia until 2013, but became super-value between 2014 and 2018, and then a value brand in 2020 after BATA introduced Rothmans Cool Crush as a super-value cigarette in 2018. The leading brand Winfield, which was mainstream until 2018, also launched Winfield Optimum and Winfield Optimum Crush with relatively low RRPs among the brand family, but also Winfield Jets and Winfield Flow Filters, as premium cigarettes in 2020. This is consistent with our finding of an increase in the proportion of cigarettes purchased that were premium brands in 2020 (Supplemental Table 1-A).

Our study revealed the average price per cigarette was consistently lower when purchased in large-size packs than in small or mid-sized packs, but the combination of pack size and market segment reduced the prices paid for small packs. A wide range of pack sizes were reported by our sample. This phenomenon of large variation in FMC pack sizes marketed in Australia has been noted previously.(55) Following implementation of plain packaging in 2012, small packs with the appearance of “bonus cigarettes” (i.e., packs of 21 or 26 cigarettes) were introduced, particularly among value and super-value brands (e.g., Bond Street 26s). In addition to offering small packs, the RRP per stick of large packs (40s or 50s) of cheap brands was 15–20% less than the market-leading pack (Winfield 25s) leading up to 2020,(2) suggesting both small and large packs are important industry strategies in Australia. The proliferation in pack/pouch sizes confuses price signals for consumers and may mitigate the impact of tax increases by both offering cheaper cigarettes when purchased in bulk and also packs with a lower upfront cost.(56,57)

Our findings support recently announced plans to mandate a standard pack/pouch size.(58) This policy would counter this marketing strategy that targets price-sensitive consumers. Aotearoa/New Zealand adopted a minimum and maximum total quantity of cigarettes per pack and RYO tobacco per pouch as part of its plain packaging legislation,(59) a single standard pack size would also simplify comparisons across products. In addition, direct price controls, such as a minimum and/or maximum prices, could be applied(57) to reduce potential cross-subsidisation to maintain low priced products and also limit the total profit available to the industry. Prohibiting discounts for bulk purchasing (e.g., buying in cartons) should also be considered. Limiting sales to two packs per transaction has also been suggested to reduce consumption.(57)

Strengths and limitations

This study has several limitations. First, potential measurement error in the outcome and predictor variables may have occurred due to reliance on self-report. Second, due to few RYO tobacco purchases, we could not assess the association between pouch size and the aforementioned factors. Third, instead of weighing rolled cigarettes, we used self-reported information to calculate RYO cigarette weights, which may produce rounding errors that may be greater for those using smaller pouches. Fourth, the average price paid by respondents for some combinations, such as FM-carton and large or small sizes, or FMC-carton and economy brand, should be cautiously interpreted due to the small sample sizes for these combinations. Finally, this study only examined tobacco purchasing behaviours and does not examine smoking reduction or use of e-cigarettes. E-cigarettes are taxed less than tobacco cigarettes.(60) Since the retail sale of e-cigarettes containing nicotine has been illegal in Australia,(61) the prevalence of e-cigarette use was low over most of the time period of this study precluding examining their use as a cost-minimising strategy, Because e-cigarettes are cheaper than cigarettes in Australia,(62) and e-cigarette use is now increasing despite regulatory barriers, future studies should include switching to or partial substitution with e-cigarettes as a cost-minimising behaviour.(63)

Despite the above limitations, to the best of our knowledge, this is the first study to examine various tobacco purchasing behaviours in the unique Australian tobacco market across a long period of time, covering the effects of plain packaging policy and both a large one-off tax increase in 2010 and a series of annual tobacco tax increases since 2013. The ITC Project’s detailed data on the prices paid by respondents and products purchased allowed us to compare the average prices paid for each outcome and to examine the association between socio-economic and smoking-related factors and purchasing behaviours. In terms of cigarette brand market segment, although the responses were self-reported, we used the market RRPs and considered brand extensions to calculate more accurate average RRPs and categorised brands into appropriate segments. We also considered a function of pack and pouch sizes with brand market segments to explore several cost-minimising purchasing options. Our findings add to the body of evidence on the impact of tobacco tax increases on tobacco purchasing behaviours, in addition to previous routine surveillance of the Australian tobacco market.

CONCLUSION

Large pack size, economy brands, and products with a low cost upfront (e.g., small RYO pouches) are more likely to be purchased by consumers with low-incomes and higher nicotine dependence. Our research identified an upward trend in purchasing cheaper tobacco products, which may be explained by tobacco industry marketing of a wide range of pack sizes, new economy brands and variants within a brand family that stretch across different market segments. Standardising pack and pouch size may reduce price-related marketing and especially benefit people on a low income or with high nicotine dependence. Establishing floor and/or ceiling prices, banning discounts for bulk purchases and limiting the number of packs sold in any transaction are additional policies that could be considered to limit industry marketing that counteracts the impact of tobacco tax increases.

What is already known on this topic

• Tobacco tax is a cost-effective intervention to prevent and reduce the prevalence of tobacco use by increasing the price and thus reducing consumer demand; However, use of cost-minimising behaviours can reduce the impact of tax increases.

• Consumers shift to cheaper tobacco products to reduce cost; Following a large tax increase in 2010, Australian consumers shifted to less expensive RYO, reducing the impact of the tax increase on smoking cessation among lower-income groups.

• Tobacco companies have also introduced unusual pack sizes and brand extensions in both factory-made cigarettes (FMC) and roll-your-own tobacco (RYO) into the Australian tobacco market to counteract the impact of tobacco tax increases.

What this study adds

• This study documents tobacco product purchases and prices paid in Australia over a period which included nine large tax increases and the introduction of plain packaging, identifying a significant increase in cost-minimising purchasing patterns (RYO, large-sized pack, and economy brand shifts from large to small size economy packs).

• As a positive result of the harmonisation of RYO tax with FMC tax starting in September 2017, the gap between the reported average FMC-RYO tobacco prices per stick (0.7g) was reduced from 24% in 2018 to 9% in 2020. However, consumers rolled thinner RYO cigarettes over the study period in response.

• Those who had a lower income, were highly addicted to nicotine and had no intention to quit smoking were more likely to purchase RYO and large-sized packs.

How this study might affect research, practice or policy

• Tobacco product pricing plays an important role in determining smoking patterns and inequalities.

• Standardising pack and pouch sizes may reduce price-related marketing (e.g., ‘bulk’ buying discounts, the appearance of ‘bonus’ cigarettes, etc) and especially benefit people on a low income and those who are highly addicted to nicotine.

• In addition to increasing tobacco tax and standardising pack and pouch sizes, effectiveness of tobacco tax policy might be increased by measures such as a minimum ‘floor’ price and/or a price ‘ceiling’.