2.6 Comparisons of quality and results using various estimates of tobacco consumption in Australia

Show / hide chapter menu

Each of the alternative sources of data used to estimate tobacco consumption—production and trade data, tax receipts, sales data, self-reported use of cigarettes and self-reported expenditure on tobacco products—has advantages and each has disadvantages. And readers will have noted that each provides different estimates of total and per capita consumption.

2.6.1 Limitations of data

While quantifying the number of cigarettes produced in Australian factories or levied for duty prior to sale may seem like a straightforward, highly objective process, it must be remembered that data on manufacturing and duties is generated by individuals interpreting and reporting on data entered into electronic databases by other individuals. There is still room for error and inconsistency over time and between individuals in the way that products are coded and the ways that quantities are calculated and aggregated.

One can be reasonably confident about the accuracy of data provided by Australian tobacco companies to the Australian Bureau of Statistics which is covered by an act of law73 requiring companies to comply with official requests. International research agencies compiling data on numbers of cigarettes produced or sold by contrast have no such legal sway over respondents, and no legislative requirement to disclose and correct errors if these emerge.

Production and trade data may be lower or higher from year to year not because retail sales have decreased, but rather because of changes in timing of production schedules, importing and exporting opportunities and warehousing practices. Some commentators suspect that tobacco companies may even alter production schedules to reduce or increase apparent production over particular periods to attempt to persuade governments that certain tobacco control initiatives are ineffective.35

Data on weight of tobacco products manufactured or excised over the years provide only a rough estimate of the numbers of cigarettes consumed given that cigarettes weights have declined over time, with little information available about the average weight in each year. Estimates of cigarette imports to and exports from Australia based on weight may somewhat underestimate the actual numbers of cigarettes being imported and exported if these are based on international rather than Australian averages of cigarette weight.

And it is important to remember that the quantities of tobacco products on which duties are levied and the quantities of tobacco products sold by licensed tobacco companies and wholesalers underestimate consumption to the extent that they miss illicit tobacco and contraband cigarettes.

Self-reports of amounts of tobacco consumed seem to significantly underestimate consumption. Some of the disparity between the figures based on official records and self-report data may also be accounted for by stock that is past used-by date or damaged and returned not sold. The estimate of total annual consumption built up from self-report data included in Table 2.10 does not take into account cigarettes smoked by children under 14 years of age. On the other hand, estimates of tobacco sales and consumption based on excise and customs receipt do not include products on which such duties has been evaded. It is clear that smokers must significantly underestimate the amount they smoke each day. There is no evidence, however, that the tendency or extent of underreporting has changed over recent times.

Industry sales data found periodically on the websites of tobacco companies or published in trade magazines should also be interpreted with caution. Methods of collecting the data and sources are often not reported. Sometimes such data are based on raw figures, sometimes they represents 12-month running averages. Often the basis for estimates is not reported. For instance, researchers rarely explain how they estimate cigarette numbers from data based on weight and vice versa. Not infrequently, figures are revised without explanation, and without corresponding revision of historical figures no longer included in data tables.

Table 2.23 attempts to compare and contrast the various limitations associated with each source of data relevant for estimating tobacco consumption in Australia.

Frustration about lack of reliable data on tobacco consumption has led researchers and policy experts to call for the government to require tobacco companies to report on sales using their own records of stock supplied to and returned from wholesalers and retailers.74, 75 Companies could be required to report quarterly and annually on sales of various categories of tobacco products on a regional, state and national basis. In this way health authorities could judge the relative success of tobacco control strategies in each jurisdiction. Such reporting of sales is required under the Smokefree Environments Act 2000 in New Zealand76, the Tobacco Reporting Regulations in Canada77 and, by regulation, to the Federal Trade Commission in the United States.

All the limitations described above warn against unqualified acceptance of any one figure as a definitive estimate of consumption in Australia. Interpreting changes over time and comparisons between countries is especially fraught.

2.6.2 Consistency of changes in various data sets

Figure 2.12 sets out per capita estimates based on all the various sources of data that provide insights into tobacco consumption in Australia.

Although the various estimates of consumption are calculated from very different data sources, the overall trends—the steepness of decline in particular periods—are surprisingly similar.

While we might not be able to say exactly what current tobacco consumption actually is at present in Australia from the similarity of the pattern of reductions in all the data sources, we can be certain that it has been reducing.

Figure 2.13 plots per capita consumption against smoking prevalence between 1980 and 2004.

Trends in our best estimates of per capita consumption of tobacco products seemed to closely mirror trends in population smoking prevalence until 1991. Between 1992 and 2001, consumption appears to have fallen more steeply than prevalence. Between 2001 and 2004, consumption per capita appeared to reduce less than prevalence did.