The Impact of China National Tobacco Company’s ‘Premiumization’ Strategy: Longitudinal Findings from the ITC China Project (2006-2015)

Abstract

Background:

In 2009, the China National Tobacco Company (CNTC) began their Premiumization Strategy, designed to encourage smokers to trade-up to more expensive brands, mainly by promoting the concept that higher class cigarettes are better quality and less harmful. This study is the first evaluation of the strategy’s impact on: (1) prevalence of premium (PBCs), mid-priced (MBCs), and discount (DBCs) brand cigarettes over 9 years, from 3 years pre-strategy (2006) to 6 years post-strategy (2015); and (2) changes in reasons for choosing PBCs, MBCs, and DBCs.

Methods:

A representative cohort of adult Chinese smokers (N= 9,047) in 7 cities who participated in five waves of the ITC China Survey: pre-implementation: [Waves: 1 (2006; n=3,452), 2 (2007–08; n=3,586)]; mid-implementation [Wave 3 (2009; n= 4,172)]; and post- implementation [Waves 4 (2011–12; n= 4,070), 5 (2013–2015; n=2,775)]. Generalized estimating equations were conducted to examine changes in prevalence of PBCs, MBCs, and DBCs, and reasons for brand choice from pre-implementation to post-implementation.

Results:

From pre- to post-implementation, there was an increase in prevalence of PBCs (5.4% to 23.2%,p<0.001) and MBCs (40.0% to 50.4%,p<0.001), and a decrease in DBCs (54.6% to 26.5%,p<0.001). There was an increase in smokers who chose their current brand because they believed it to be less harmful, both for MBC smokers (+13.0%,p=0.001), and PBC smokers (+9.0%,p=0.06). There was an increase for smokers in all brand classes for choosing their current brand because they were ‘higher in quality’ and because of affordable price, but the greatest increase was among PBC smokers (+18.6%, p<0.001 and +34.9%,p<0.001, respectively).

Conclusions:

Our findings demonstrate that the rising trend in Chinese smokers’ choice of ‘less harmful’, ‘higher quality’, and ‘affordable’ cigarettes, particularly PBCs is likely due to CNTC’s aggressive marketing strategies. Strong tobacco control policies that prohibit CNTC’s marketing activities are critical in order dispel erroneous beliefs that sustain continued smoking in China, where the global tobacco epidemic is exerting its greatest toll.

INTRODUCTION

The tobacco epidemic in China is a public health challenge of unmatched proportions.¹ More than 300 million people in China smoke (including about half of all men), which represents one–third of the world’s smokers, and approximately 1 million tobacco–attributable deaths occur every year.² China’s cigarette market is vast, with a total of 2.4 trillion cigarettes consumed each year. Chinese smokers consume more cigarettes than smokers in all other low– and middle–income countries combined.³ Without effective measures to reduce tobacco use, the number of annual tobacco–related deaths in China is projected to reach 3 million by 2050.⁴

The China National Tobacco Company (CNTC) is a state–owned enterprise, with a monopoly of the cigarette market, accounting for 98% of domestic sales.^5–7 In 2015, CNTC’s gross profit was 303 billion Chinese Yuan Renminbi (RMB) (about 48 billion USD),⁸ making it the most profitable company in the country. The tobacco industry contributed 840.4 billion RMB (equivalent to about 122 billion USD) in tobacco tax revenue in 2015, a 9% increase from their contribution in 2014. In all, CNTC contributes 7–10% of Chinese central government’s total annual revenue through tobacco tax and profit–sharing,^3,7 not including revenues shared to local governments.

Switching Between Cigarette Brands in China in Response to Price Increases

Relative to other countries where the variability of cigarettes prices is smaller, Chinese smokers offset price increases by switching to cheaper cigarette brands (and which are often still in the same price class).⁹ White et al. (2014) found that although a substantial number (38%) of smokers switched price tiers between waves of the ITC China Survey, about 1–1.5 years apart, more than half switched brands between waves but stayed within the same price tier.⁹ Interestingly, trading up was more common among smokers in mid-priced classes, which likely reflected rising affordability and only a nominal 1.5% price increase between 2006 to 2009. This demonstrates that Chinese smokers are relatively flexible in brand choices and do not generally display strong loyalty to one brand variety.

CNTC’s Cigarette Premiumization Strategy

Historically, the majority of cigarette sales in China have been discount brand cigarettes (DBCs) and mid-priced brands (MBCs). However, as China experienced annual double-digit economic growth in the early 2000s,¹⁰ resulting in a surge in disposable household income, Chinese smokers could afford to spend more on cigarettes,^7,11–13 which ultimately would have fueled an increase in cigarette consumption, and discouraged smoking cessation.¹⁴

The increasing affordability created a lucrative opportunity for CNTC to increase revenue and profit. Beginning in the mid-2000s, CNTC began to adjust their cigarette brand portfolio, starting first with two important actions. First, CNTC reduced the 1000+ brand and brand variants through elimination and consolidation into a much smaller number of brands to increase production efficiency, and to increase product appeal to consumers in both domestic and foreign markets.⁷ Second, they changed the price structure of its brand portfolio (“Juan Yan Jie Gou Ti Sheng”) by decreasing sales of cheaper DBCs and increasing sales of high-end cigarettes by encouraging DBC smokers to trade up to MPCs or PBCs, and MBC smokers to trade up to PBCs. To achieve their “trading up” goal, in January 2009, CNTC introduced their 5-year Premiumization Strategy (2010–2015).^7,8,15,16 This action plan resulted in a sharp increase in the number of new cigarette brand variants that entered the domestic market (particularly in the post-Premiumization (post-implementation) period, see Supplemental Figure 1). Between 2012 and the first half of 2017, 615 new cigarette brand variants entered China’s tobacco market, where 510 (83%) were PBCs, 82 (13%) were MBCs, and only 23 (4%) were DBCs¹⁷ (Supplemental Figure 2).

While marketing PBCs by emphasizing their superiority, quality, luxury, and exclusivity, CNTC also linked PBCs to guanxi—a Chinese system of social networks and influential relationships to facilitate business and other dealings,^7,18 and positioning PBCs as ideal for gift-giving. Cigarette gifting and sharing are important and common Chinese social networking practices (primarily an exclusive practice among men), and has been shown to have a significant influence on brand preference.¹⁹ These social practices not only drive cigarette consumption among established smokers,²⁰ but have been linked to increased smoking initiation, failure to quit smoking,^20–22 and increased secondhand smoke at home.²³ Gifting and sharing have significantly contribute to the smoking epidemic in China.^19,20

To our knowledge, this is the first study to evaluate the impact of CNTC’s Premiumization Strategy. We conducted a longitudinal analysis over a 9-year period (2006–2015) starting in the pre-implementation period, through the mid-implementation period, and finally in the post-implementation period. Our longitudinal analyses examined: (1) changes in prevalence of the three cigarette brand categories (PBCs, MBCs, DBCs); (2) the pre- and post-implementation changes in reasons for choosing PBCs, MBCs, and DBCs, including those reasons that were specifically targeted in the CNTC strategy (perceptions of quality and of harmfulness) and those that were not specifically targeted (e.g., greater affordability); and (3) predictors of choosing PBCs and MBCs compared to DBCs.

METHODS

The International Tobacco Control (ITC) China Survey is a longitudinal prospective cohort study of a representative sample of adult (≥18 years) daily and weekly smokers. Five survey waves were conducted between 2006 and 2015 from six cities in Waves 1 and 2 (Beijing, Changsha, Guangzhou, Shanghai, Shenyang, Yinchuan), seven cities in Waves 3 and 4 (Beijing, Changsha, Guangzhou, Shanghai, Shenyang, Yinchuan, Kunming), and five cities in Wave 5 (Beijing, Guangzhou, Shanghai, Shenyang, Kunming).

At Wave 1, a multistage cluster sampling design was used to randomly identify smokers through household enumeration in each of the respective cities. Informed written consent was obtained, and the survey was conducted by face-to-face interview (approximately 45–60 minutes). To maintain the sample size over time, cohort members lost to follow-up at each wave were replenished by newly recruited individuals from the same sampling frame as Wave 1.²⁴ Respondents were eligible for this study if they had completed all survey questions regarding their cigarette brand.

For this evaluation study, the five waves were categorized into three periods: Pre-implementation, consisting of Wave 1 (April-August 2006) and Wave 2 (October 2007-January 2008); Mid-implementation, consisting of Wave 3 (May-October 2009); and Post-implementation, consisting of Wave 4 (September 2011-November 2012), and Wave 5 (November 2013-July 2015).

Measures

CNTC’s Cigarette Price Classification System:

The structure of CNTC’s internal classification system of PBCs, MBCs, and DBCs arises from a clustering of five cigarette classes (or grades) based on before-tax allocation price: PBCs consist of Class I (>10 RMB/pack) and Class II (7–10 RMB/pack); MBCs consist of Class III (3–7 RMB/pack); and DBCs consist of Class IV (1.65–3 RMB/pack) and Class V (<1.65 RMB/pack).

Current regular brand choice:

Each respondent’s current regular brand choice was assessed with this question: “in the last 30 days, what brand of cigarettes did you smoke more than any other?” Respondents were asked to show their cigarette package and the barcode was recorded by the interviewer. The barcode was used by the research team to determine the brand and brand variant using CNTC’s cigarette catalog, the official list of all CNTC brands and brand varieties (accounting for 98% of domestic sales).²⁵ Each brand variety is listed in the catalog with both a photograph of the package. This information was used to classify smokers’ brand explicitly and unambiguously into the five classes (I through V) using the prices for each brand that were gathered from CNTC sources, and then subsequently into the broader categories of PBC, MBC, or DBC using CNTC’s cigarette price classification system. This matching process was conducted by the first author (SX) and by a research assistant (both of whom are native Chinese speakers).

Reasons for brand selection:

Reasons for brand selection were measured by answering ‘yes’ or ‘no’ to four questions: “In choosing your current regular brand was part of your decision to smoke this brand based on: they are less harmful to your health, affordable price, high quality, or received as a gift.

Demographic and socio-economic variables:

Demographic variables were: sex, age (18–24, 25–39, 40–54, 55+), and monthly household income (less than 1000 Yuan, 1000–3000 Yuan, 3000–5000 Yuan, 5000–7000 Yuan, 7000–9000 Yuan, and 9000+ Yuan; One Yuan = 0.16 USD).

Cigarette affordability was defined as the ratio of ‘cigarettes price per day’ to ‘household per-capita daily income’. Cigarette price per day was calculated by multiplying the respondent’s self-reported cigarette price per stick (converted from the price they paid for their pack of cigarettes) and the number of cigarettes smoked per day (consumption). Household per-capita daily income was calculated by dividing household monthly income by the average number of days in a month (30.4) and then by the number of adults in the household. For Waves 1 to 4, where household income was given in categories, the mid-point at each income category was used for the calculation. This index is unitless and is comparable between respondents (and represents the proportion of personal income spent on cigarettes). It is a modified version of the relative income price measure of affordability that was first used by Blecher & van Walbeek (2004),²⁶ modified in that it was computed at the individual level. A lower ratio is indicative of higher affordability.

Smoking status:

Current smokers self-reported having smoked at least 100 cigarettes, and smoked every day (daily smoker) or somedays (non–daily smokers).

Time in sample:

In longitudinal surveys, individuals’ responses may differ depending on how many waves (surveys) they have participated in. Therefore, it is necessary to control for these time-in-sample (TIS) effects by adding a value for each respondent to all analytic models. Including a TIS variable [which is equal to the total number of waves that the respondent was present (e.g., a respondent who entered the study at Wave 2, and also participated at Waves 3, 4 and 5, would be given a numeric TIS value of 4] is a standard procedure for all longitudinal analyses of ITC survey data.²⁷

Statistical Analyses

Initial unweighted descriptive statistics were used to describe respondent characteristics at baseline (see Table 1).

Table 1.

Sample Characteristics of Respondents at Recruitment by Wave.

Characteristics		Wave 1 (2006) (n=3452 )		Wave 2 (2007–08) (n=3586 )		Wave 3 (2009) (n=4172)		Wave 4 (2011–12) (n=4070)		Wave (2013–15) (n=2775)		Total (N=9,047)
		n	%	n	%	n	%	n	%	n	%	n	%
Age group	18–24	40	1.2	33	0.9	75	1.8	58	1.4	40	1.4	151	1.7
	25–39	604	17.5	588	16.4	807	19.3	730	18.0	462	16.7	1678	18.6
	40–54	1708	49.5	1811	50.5	1970	47.2	1837	45.2	1148	41.4	4025	44.5
	55+	1100	31.9	1154	32.2	1320	31.6	1442	35.5	1121	40.5	3186	35.2
Sex	male	3293	95.4	3411	95.1	3977	95.3	3888	95.5	2627	94.7	8841	95.0
	female	159	4.6	175	4.9	195	4.7	182	4.5	148	5.3	465	5.0
City	Beijing	521	15.1	608	17.0	586	14.1	531	13.1	517	18.6	1238	13.7
	Shenyang	401	11.6	553	15.4	619	14.8	570	14.0	510	18.4	1488	16.5
	Shanghai	708	20.5	728	20.3	670	16.1	675	16.6	665	24.0	1370	15.1
	Changsha	765	22.2	666	18.6	635	15.2	632	15.5	0	0.0	1260	13.9
	Guangzhou	532	15.4	433	12.1	346	8.3	326	8.0	419	15.1	1112	12.3
	Kunming	0	0.0	0	0.0	724	17.4	678	16.7	664	23.9	1259	13.9
	Yinchuan	525	15.2	598	16.7	592	14.2	658	16.2	0	0.0	1320	14.6
Income	<100 Yuan	652	18.9	591	16.5	404	9.7	213	5.2	35	1.3	662	7.3
	1000–2999 Yuan	1548	44.8	1622	45.2	1609	38.6	1164	28.6	299	10.8	2644	29.2
	3000–4999 Yuan	668	19.4	746	20.8	1188	28.5	1257	30.9	666	24.0	2360	26.1
	5000–6999 Yuan	234	6.8	269	7.5	465	11.2	678	16.7	730	26.3	1439	15.9
	7000–8999 Yuan	57	1.7	61	1.7	134	3.2	298	7.3	338	12.2	578	6.4
	9000 Yuan or above	59	1.7	71	2.0	135	3.2	284	7.0	491	17.7	771	8.5
	Don’t Know	234	6.8	226	6.3	237	5.7	176	4.3	216	7.8	593	6.6
Smoking status	Daily	3250	94.2	3420	95.4	3996	95.8	3863	94.9	2627	94.7	8533	94.3
	Weekly	202	5.9	166	4.6	176	4.2	206	5.1	148	5.3	513	5.7
Cohort	recruited in Wave 1	3452	100.0	2943	82.1	2302	55.2	1524	37.4	737	26.6	4336	47.9
	recruited in Wave 2	-	-	643	17.9	476	11.4	251	6.2	130	4.7	784	8.7
	recruited in Wave 3	-	-	-	-	1394	33.4	789	19.4	448	16.1	1516	16.8
	recruited in Wave 4	-	-	-	-	-	-	1506	37.0	665	24.0	1616	17.9
	recruited in Wave 5	-	-	-	-	-	-	-	-	795	28.7	795	8.8

Weighted prevalence estimates of PBC, MBC, and DBC use, and reasons for their current regular brand choice, were calculated at each wave, and also for the pre- and post-implementation periods using logistic regression analyses incorporating generalized estimating equations (GEE).²⁸ Demographic and socio-economic explanatory variables and reasons for PBC or MBC brand choice (compared to DBCs) were fitted using GEE models with the multinomial logit link. Pre- and post-implementation differences between PBC, MBC, and DBC smokers for the reasons for choosing their current regular brand were also modeled and tested using GEE (with the logit link). All analyses controlled for sex, age, city, cigarette affordability, smoking status and TIS.

To address a potential design effect resulting from the complex longitudinal survey design and within-individual correlations due to repeated measures at each wave, a nested structure that includes the strata (cities), the primary sampling units (neighbourhood or Ju Wei Hui) and the respondent IDs (with repeated measures at each wave) was used to construct the models. The rescaled cross-sectional weights at recruitment were applied to each respondent for modeling. The technical details on weight calculation are available in the ITC China Project technical reports: http://www.itcproject.org/technical-report/?country=China.

All the above analyses were conducted using SAS–callable SUDAAN (Version 11).²⁹ The predicted marginal standardization method in the SUDAAN GEE model (PREDMARG) was used for estimating prevalence.³⁰ General linear contrasts of the predicted marginals in the corresponding models were specified to test the significance of between-wave percent changes. All confidence intervals (CI) and statistical significance were tested at the 95% confidence level.

RESULTS

Brand Code Eligibility

Across the 5 waves, there were 23,084 smoking respondent records. Of these, 94.6% (21,841 records) provided cigarette package barcodes to the interviewer, leaving 1,243 missing records. Among the 21,841 respondent records over the 5 waves who provided a bar code, 87% (19,011 records) could be assigned to a brand family and variant. Among the 19,011 records with correctly identified brand names and varieties, 95% (18,055 records) could be classified by price; these were categorized into one of CNTC’s 5 price classes. The end result of this process was that 78.2% (18,055/23,084) of the original records were valid for inclusion in this study.

Study Sample

The current study included smokers (daily and non-daily) for which information about their current regular cigarette brand could be ascertained at the time of the survey interview (see description above). Of the 10,201 current daily or weekly smokers who participated in any of the five wave surveys, 9,047 had complete survey data and their current regular cigarette brand choice was validated, and thus were included in all subsequent analyses.

Table 1 presents the demographic and smoking behaviour characteristics of the sample at recruitment. In brief, the average age of the sample at the time of recruitment was 50.3 years. Nearly all of the respondents were male (95%) and daily smokers (94%).

Prevalence Trends of the Three Cigarette Brand Choices Over the 9-Year Period.

Between 2006 and 2015, there was an increase in PBCs (4.7% to 32.1%) and MBCs (32.3% to 48.1%), but a substantial decrease in DBCs (57.0% to 19.8%) (see Table 2). In particular, the largest changes for all three brand classes occurred during the post-implementation period, especially for PBCs where there was a greater influx of new PBCs entering the market.

Table 2.

Prevalence of PBCs, MBCs and DBCs From Pre-implementation (2006) to Post-implementation (2013–15) of the Premiumization Strategy.

Wave (Survey Year)	Implementation Timing	Premium Brands (PBCs)		Mid-Priced Brands (MBCs)		Discount Brands (DBCs)
		%	(95% CI)	%	(95% CI)	%	(95% CI)
Wave 1 (2006)	Pre- implementation	4.7	(3.4 – 6.3)	38.3	(34.0 – 42.9)	57.0	(52.2 – 61.7)
Wave 2 (2007–2008)	Pre- implementation	6.1	(4.5 – 8.2)	41.8	(37.6 – 46.1)	52.1	(47.8 – 56.3)
Wave 3 (2009)	Mid- implementation	8.8	(7.2 – 10.6)	44.0	(40.0 – 48.1)	47.2	(42.6 – 51.9)
Wave 4 (2011–2012)	Post- implementation	19.5	(16.6 – 22.7)	52.4	(47.3 – 57.4)	28.2	(22.5 – 34.6)
Wave 5 (2013–2015)	Post- implementation	32.1	(28.7 – 35.8)	48.1	(43.2 – 53.0)	19.8	(14.7– 26.1)

Notes: the prevalence estimates controlled for sex, age, time in sample, city, affordability, and smoking status.

Data are weighted and standardized.

As shown in Figure 1, there was a significant increase for PBCs (5.4% to 23.2%, p<0.001) and MBCs (40.0% to 50.4%, p<0.001), and a significant decrease for DBCs (54.6% to 26.5%, p<0.001) between pre-and post-implementation. The increase for PBCs (+27.4%) was greater than the increase for MBCs (+9.8%, p<0.001) between pre-and post-implementation.

Figure 1.

Prevalence of Brand Choice Pre- and Post Premiumization Stategy among Urban Adult Smokers in China

Reasons for Current Brand Choice between Pre- and Post- implementation

Current Brand Choice is Less Harmful to Your Health

As shown in Table 3, there was an increase for smokers in each of the brand classes for choosing their current brand because they were less harmful for their health, although the change was only significant for MBCs (p=0.001), and was marginally significant for PBCs (p=0.06). The difference was not significant between DBCs (+ 4.4% increase) and PBCs (+ 9.0% increase) or MBCs (+13.0% increase).

Table 3.

Reasons for Smokers Choosing their Current Cigarette Brand: Before and After CNTCs Premiumization Strategy.

	Premium Brands (PBCs)				Mid-Priced Brands (MBCs)				Discount Brands (DBCs)
	n	%	SE	(95% CI)	n	%	SE	(95% CI)	n	%	SE	(95% CI)
Less Harmful
Pre-implementation	146	46.2	3.9	38.6 – 53.9	940	39.3	2.6	34.1 – 44.7	1261	35.1	2.5	30.3 – 40.2
Post- implementation	831	55.2	2.3	50.5 – 59.7	1807	52.3	2.3	47.7 – 57.0	742	39.5	3.0	33.6 – 45.7
Pre-post % Difference		+ 9.0%				+ 13.0%				+ 4.4%
p-value of pre-post difference		0.06				0.001				0.26
p-value of pre-post difference of PBCs vs. DBCs and MBCs vs. DBCs		0.39				0.07				—
Higher in Quality
Pre-implementation	247	68.3	4.2	59.5 – 76.0	1670	65.2	2.7	59.6 – 70.4	2311	58.8	2.9	52.9 – 64.4
Post- implementation	1387	86.9	1.5	83.5 – 89.7	2829	81.3	1.6	77.8 – 84.4	1070	67.2	2.5	62.1 – 71.9
Pre-post % Difference		+ 18.6				+ 16.1				+ 8.4%
p-value of pre-post difference		0.0002				<0.0001				0.012
p-value of pre-post difference of PBCs vs. DBCs and MBCs vs. DBCs		0.036				0.025				―
Affordable Price
Pre-implementation	162	46.9	5.2	36.9 – 57.2	1660	63.5	2.6	58.1 – 68.6	2979	79.9	1.4	77.1 – 82.5
Post- implementation	1403	81.8	2.5	76.2 – 86.3	3209	90.0	1.2	87.2 – 92.2	1405	92.9	1.6	89.1 – 95.5
Pre-post % Difference		+ 34.9				+ 26.5				+ 13.0
p-value of pre-post difference		<0.0001				<0.0001				<0.0001
p-value of pre-post difference of PBCs vs. DBCs and MBCs vs. DBCs		0.0001				<0.0001				―

Current Brand Choice is Higher in Quality

There was a significant increase for smokers in each of the brand classes for choosing their current brand because they were higher in quality. PBC smokers had the greatest increase (+18.6, p=0.0002) followed by MBCs (+16.1%, p<0.0001) and DBCs (+8.4%, p=0.012). PCB smokers and MBC smokers had a greater increase than DBC smokers in choosing their brand because of higher quality (Table 3).

Current Brand Choice is More Affordable

There was a significant increase for smokers in all three brand classes for choosing their current brand because of affordable price. But PBC smokers had the largest increase in choosing their current brand because of price (+34.6%, <0.0001), followed by MBC smokers (+26.5%, <0.0001) and DBC smokers (+13.0%, <0.0001). There was a greater increase over time for PBC (p=0.0001) and MBC (p<0.0001) smokers to choose their brand because of price compared to DBC smokers (Table 3).

Predictors of PBC and MBC Use Compared to DBCs

Demographic and Socio-economic Predictors of Current Brand Choice

PBC smokers were more likely to be: younger (with the greatest use among 25–39 year olds compared to the oldest age group), male, from Shanghai, Kunming, or Yinchuan, a non-daily smoker, and have greater cigarette affordability (see Table 4a). MBC smokers were also more likely to be younger in age, from all other cities compared to Beijing, and less likely to be a daily smoker. And although MBC smokers had significantly greater cigarette affordability compared to DBC smokers, this was more apparent among the PBC group.

Table 4a.

Demographic and Socio-economic Predictors of PBCs and MBCs as a Regular Brand Compared to DBCs.

			Premium brand (PBC) smoker		Mid-price brand (MBC) smoker
Explanatory variables	n	% n (raw)	Odds Ratio (95% CI )	P–value	Odds Ratio (95% CI )	P–value
Intercept			0.01 ( 0.00 – 0.02)	<0.001	0.14 ( 0.07 – 0.26)	<0.001
Survey waves
Wave 1	3452	19.1	Reference		Reference
Wave 2	3586	19.9	1.50 ( 0.87 – 2.58)	0.1	1.39 ( 1.09 – 1.78)	0.009
Wave 3	4172	23.1	2.89 (1.91 – 4.37)	<0.001	1.68 ( 1.27 – 2.22)	<0.001
Wave 4	4070	22.5	16.10 (9.85 – 26.32)	<0.001	4.47 (3.10 – 6.44)	<0.001
Wave 5	2775	15.4	51.66 (29.34 – 90.95)	<0.001	7.01 ( 4.28 – 11.46)	<0.001
Age group
18–24	246	1.4	5.41 ( 2.24 – 13.03)	<0.001	3.26 ( 1.52 – 7.03)	0.003
25–39	3191	17.7	7.94 ( 5.44 – 11.60)	<0.001	3.56 ( 2.89 – 4.38)	<0.001
40–54	8474	47.0	2.66 ( 1.96 – 3.61)	<0.001	1.79 ( 1.51 – 2.12)	<0.001
55+	6137	34	Reference		Reference
Sex
Male	17196	95.24	2.07 ( 1.03 – 4.17)	0.041	1.34 ( 0.97 – 1.84)	0.076
Female	859	4.76	Reference		Reference
City
Beijing	2763	15.3	Reference		Reference
Shenyang	2653	14.7	1.19 ( 0.65 – 2.20)	0.567	1.73 ( 1.05 – 2.84)	0.032
Shanghai	3446	19.1	20.41 (11.71 – 35.57)	<0.001	17.80 (11.54 – 27.46)	<0.001
Changsha	2698	14.9	0.82 ( 0.42 – 1.60)	0.564	0.51 ( 0.30 – 0.86)	0.012
Guangzhou	2056	11.4	0.74 ( 0.35 – 1.56)	0.429	2.74 ( 1.49 – 5.05)	0.002
Kunming	2066	11.4	14.65 ( 6.94 – 30.90)	<0.001	7.89 ( 4.59 – 13.54)	<0.001
Yinchuan	2373	13.1	2.85 ( 1.40 – 5.80)	0.004	2.38 ( 1.54 – 3.67)	<0.001
Time in sample
1	7802	43.2	1.67 ( 0.89 – 3.13)	0.111	1.30 ( 0.69 – 2.43)	0.414
2	4940	27.4	1.35 ( 0.68 – 2.69)	0.391	1.21 ( 0.64 – 2.27)	0.558
3	2988	16.6	1.11 ( 0.66 – 1.85)	0.694	1.19 ( 0.69 – 2.04)	0.522
4	1631	9.0	1.06 ( 0.65 – 1.72)	0.821	0.91 ( 0.60 – 1.39)	0.671
5	694	3.8	Reference		Reference
Smoking status
Daily	17156	95.0	0.29 ( 0.20 – 0.43)	<0.001	0.65 ( 0.48 – 0.89)	0.007
Weekly	898	5.0	Reference		Reference
Affordability			6.52 ( 4.25 −10.00)	<0.001	1.62 ( 1.11 – 2.36)	0.012

Note: The variables in the model are presented in Table 4b.

Next, when comparing the use of PBCs vs. DBCs over time (where Wave 1 was the reference), there was no significant increase in the odds of smokers using PBCs vs. DBCs between Waves 1 (pre-implementation) and 2 (pre-implementation; OR=1.50, p=0.1). However, the use of PBCs (vs. DBCs) was more likely to have significantly increased by mid-implementation (Wave 3, OR=2.89, p<0.001), and even more so at the post-implementation periods, Waves 4 (OR=16.10, p<0.001) and 5 (OR=51.66, p<0.001). Moreover, after controlling for all personal-level characteristics, affordability and reasons for choosing current brand, PBC use increased exponentially over time, with a much more significant increase at Wave 5 compared to Wave 4.

Finally, when comparing the use of MBCs vs. DBCs over time, there was a significant increase in the use of MBCs by Wave 2 (OR=1.39 1.09, p=0.009). And although there were significant increases over time by mid-implementation (OR=1.68, p<0.001) and post-implementation Wave 4 (OR=4.47, p<0.001) and 5 (OR=7.01, p<0.001), these increments were much less pronounced than the changes in PBC use over time (and there was little difference between Waves 4 and 5 for MBC vs. DBC use as compared to PBC vs. DBC use), thus pointing to the much greater shift from DBCs to PBCs after the implementation of the Premiumization Strategy.

Reasons as Predictors of PBC and MBC Use Compared to DBCs

Compared to DBC smokers, PBC smokers were significantly more likely to choose their brand because they were: less harmful, received as a gift from others, and higher in quality. DBC smokers were more likely to choose their brand because of affordable price (see Table 4b).

Table 4b.

Reasons that are associated with PBCs and MBCs as a Current Brand Choice Compared to DBCs.

			Premium brand (PBC) smoker		Mid-price brand (MBC) smoker
Explanatory variables	N	Raw %	Odds Ratio (95% CI )	P–value	Odds Ratio (95% CI )	P–value
Intercept			0.01 ( 0.00 – 0.02)	<0.001	0.14 ( 0.07 – 0.26)	<0.001
Reason for choosing own brand: less harmful
Yes	7372	43.2	1.42 ( 1.18 – 1.70)	<0.001	1.19 ( 1.02 – 1.40)	0.026
No	9690	56.8	Reference		Reference
Reason for choosing own brand: affordable price
Yes	14192	79.3	0.16 ( 0.12 – 0.21)	<0.001	0.39 ( 0.32 – 0.47)	<0.001
No	3710	20.7	Reference		Reference
Reason for choosing own brand: gift from others
Yes	1628	9.2	2.94 ( 1.89 – 4.58)	<0.001	1.58 ( 1.14 – 2.19)	0.007
No	16116	90.8	Reference		Reference
Reason for choosing own brand: high quality
Yes	12500	70.6	3.26 ( 2.70 – 3.95)	<0.001	2.03 ( 1.81 – 2.27)	<0.001
No	5217	29.5	Reference		Reference

Note: The other covariates in the model are presented in Table 4a.

Compared to DBC smokers, MBC smokers were more likely to choose their brand because of lower harm, they were received as a gift from others, and because they are higher in quality. DBC smokers were more likely to choose their brand because of affordable price (see Table 4b).

DISCUSSION

The present study demonstrates that CNTC’s Premiumization Strategy, coupled with increased affordability, has been successful in shifting the market from low-priced brands to the more profitable brands (PBCs and MBCs), with the greatest increase in PBCs, whose market share increased by 17.8 percentage points (a relative increase of 2.3-fold) from pre- to post-implementation.

The reasons for smokers choosing their current brand because of ‘lower harm’ increased within each of the brand classes between pre-and post-implementation (although this was only significant for MBCs, and marginally significant for PBC smokers). Additionally, there was a significant increase for smokers among all brand classes for choosing their current brand because they were ‘higher in quality’ and because of price, with the greatest increase among PBC smokers. This strongly suggests that CNTC’s Premiumization Strategy had an effect for smokers among all brand classes, but to a much greater degree for the PBCs and MBCs compared to the DBCs. This may also be evidence that CNTC took advantage of the increase in cigarette affordability, as this measure had the largest pre-post increases for all 3 brand classes.

These findings are similar to those of nationally representative surveys that have shown that Chinese smokers are susceptible to industry marketing messages about the (lack of) harmfulness of cigarette products. For example, the Global Adult Tobacco Survey found that only 10% of smokers are aware that that low tar and light cigarettes are equally as harmful as regular cigarettes.² Other ITC China studies have shown that the majority of smokers believe that ‘light’ and ‘low tar’ cigarettes are less harmful than regular cigarettes,^31,32— the highest level of over 20 ITC countries.^33,34 Gravely et al. (2017) showed that although a substantial proportion of smokers in ten low-middle income countries erroneously believed that light, low tar, and menthol cigarettes are less harmful, and that filters reduce harm, the findings were particularly disturbing in China, where prevalence of industry-induced misconceptions was the highest for all of these measures.³² The current study links those findings on the perceptions of harmfulness to CNTC’s Premiumization Strategy, which included conveying the concept that higher class cigarettes are less harmful: a higher proportion of PBC smokers (55%) at post-implementation chose their brand because they believed them to be less harmful than did MBC smokers (52%) and DBC smokers (39%). PBC smokers were 1.4 times more likely than DBC smokers to choose their brand because they believed them to be less harmful.

In addition to marketing higher class cigarettes, particularly the PBCs, as less harmful and higher quality, CNTC took advantage of another vulnerability --- the common custom of gifting and sharing of cigarettes. Cigarette smoking is part of the Chinese culture, and gifting and sharing cigarettes is a popular and normal experience, and plays an important part in social functions. Gifting of cigarettes, specifically luxury premium brands, cultivates new relationships, expresses gratitude, shows respect, and demonstrate financial success of the giver.^19,20,35 It is a social networking phenomenon that is not experienced in any other country in the world. The results from this study support that smokers’ brand preferences were influenced by receiving their brand as a gift, where three times more PBC smokers and 1.5 times more MBC smokers reported that they chose their brand because they were received as a gift compared to DBC smokers.

Strengths and limitations

The strengths of this study include the large sample size, rigorous longitudinal study design, and the ability to link smokers’ current regular cigarette brands with their reasons for choosing their brand. However, there were some limitations. One limitation is that we used the CNTC’s (before-tax) allocation price instead of the actual price paid at the last purchase to classify cigarettes as premium, mid-priced, or discount brands. CNTC allows retailers to sell cigarettes with an adjustment of 10% above or below the recommended retail price; this may have led to a small number of misclassifications if the adjustment moved a brand from one category to the adjoining one. It should be noted, however, that this variability led to decreased reliability, not validity/bias, and lower reliability leads to a lower likelihood of statistical significance. In other words, the significant differences obtained were despite the lower reliability, not because of it. Second, the measures used were self-reported, which may be subject to recall bias and social desirability, although it is not clear how this would have distorted the pattern of findings. Additionally, among smokers we do not know if when they started using premium brands at any time over the 5 waves of data if this would be their first time ever smoking premium cigarettes in their lifetime. We would therefore consider that most brand switching patterns observed in the data were due to changes in brand preferences of smokers, rather than new initiations. Finally, the composition of the ITC cohorts in China only allowed for analyses of adult smokers; we thus could not examine the effect of the Premiumization Strategy on new smokers, including youth.

CONCLUSION

The Chinese tobacco industry is a state-owned enterprise with a monopoly, and its political influence, fueled by its considerable contributions to the Chinese economy, have led to slow and ineffective governmental action in tobacco control.^33,36–39 The findings from this study are an important manifestation of this influence: the Premiumization Strategy over the past decade that has taken advantage of the expanding gap between cigarette prices (kept low in large measure because of the lack of tax increases) and smoker’s willingness to pay for “better quality” and “lower harm” products. Our findings demonstrate that the rising trend in Chinese smokers’ choice of premium brands is likely due to CNTC’s aggressive marketing strategies, coupled with greater cigarette affordability. Notably, the strong smoking-social relationship created by the tobacco industry must be broken by deglamorizing smoking and cigarette gifting. Strong tobacco control policies that prohibit CNTC’s marketing activities are critical in order dispel erroneous beliefs that sustain continued smoking in China, where the global tobacco epidemic is exerting its greatest toll.

What this paper adds.

• In early 2009, in the midst of strong economic growth, the China National Tobacco Company (CNTC) started to push forward their Premiumization Strategy, which aimed to shift the domestic tobacco market towards more expensive cigarette brands by promoting the concept that higher class cigarettes are better quality and less harmful, particularly premium brand cigarettes (PBCs).

• The present study demonstrates that CNTC’s cigarette Premiumization Strategy, coupled with increased affordability, has been successful in shifting the market from low-priced brands to the more profitable brands (PBCs and MBCs), with the greatest increase in PBCs, whose market share increased by 17.8 percentage points (a relative increase of 2.3-fold) from pre- to post-implementation. DBCs significantly decreased by 28.1% by the post-measure (a relative decrease of 51%).

• Reasons for choosing their current brand because it was believed to be ‘less harmful’ increased at the post-implementation by 13.0% for MBC smokers and by 9.0% for PBC smokers.

• There was a significant increase for smokers in all brand classes for choosing their current brand because they were ‘higher in quality’ and because of ‘affordable price’, with the greatest increase among PBC smokers for both measures (by 18.6% and 34.9% respectively).

• These findings demonstrate the need for stronger tobacco control measures in China, particularly implementing large graphic warnings, increasing the price of cigarettes, stronger advertising bans, and public health campaigns that would contribute to the denormalization of gifting cigarettes and dispel erroneous beliefs by smokers.