Estimating price elasticities when there is smuggling: the sensitivity of smoking to price in Canada
Introduction
One of the most common sources of government revenue raising around the world is the taxation of tobacco products. A central determinant of the optimal level of tobacco taxation is the price sensitivity of demand for cigarettes. This parameter has been estimated repeatedly in the US context, with a consensus estimate of 0.4–0.5 (Chaloupka and Warner, 2000), but with recent estimates at 0.6 (Gruber and Koszegi, 2000, Gruber and Koszegi, 2002; Yurekli and Zhang, 2000). But estimating this parameter in many other countries faces a fundamental difficulty which is not a real barrier in the US context: widespread smuggling. Smuggling of cigarette products can significantly bias price elasticities estimated using legal sales data, as the shift from legal to illegal product will appear to be a price sensitivity of overall demand.
A classic example of this problem is the experience of Canada in the early 1990s. Between 1989 and 1993, excise taxes at the federal and provincial level rose sharply from an average of CAN$ 1.90 to 3.50 per pack. In response to these large tax increases, there was an enormous increase in smuggling in Canada through legal export and illegal re-import. Indeed, by the typical measure of smuggling used to describe this experience, smuggled cigarettes represented roughly one-third of all domestic cigarette consumption at their peak. Then, in the face of enormous smuggling, federal and provincial taxes were halved in 1994. In recent years, however, there has again been a trend towards increasing these taxes in response to reports that the decline in cigarette prices has resulted in a significant increase in smoking by Canadians.1
In this paper, we provide a framework for estimating elasticities in the context of widespread smuggling. In particular, we estimate demand models for Canada that attempt to correct for the smuggling problem in two different ways. First, we use legal sales data, and exclude the regions and years where the smuggling problem was the worst. Second, we use micro-data on consumer cigarette expenditures. We find that the estimates from these two sources are quite divergent initially, but become more similar (in the range of −0.45 to −0.47) when smuggling provinces and years are excluded. This suggests that a fairly reliable estimate of the elasticity of demand for cigarettes can be provided for Canada despite the large smuggling problem.
We then extend the analysis in two important directions. First, we estimate the price sensitivity of demand by different income groups. It is well known that lower income groups spend a larger share of their incomes on cigarettes, so that tobacco taxes have been traditionally viewed as regressive. But, as we discuss further below, under alternative (plausibly more reasonable) models of the smoking decision, tobacco taxes may not be regressive if lower income groups are much more price sensitive. It therefore becomes critical to explore the price sensitivity of smoking by income group. We do so using our micro-data on tobacco expenditure, and show that the lowest income group is much more price sensitive than higher income groups; this finding suggests that tobacco taxes in fact may not be regressive in the context of plausible alternative models.
Second, we estimate the impact of cigarette price changes, and of smuggling of cigarettes, on demand for alcohol. If cigarettes and alcohol are substitutes, then one potential advantage of smuggling is that it reduces a shift into this alternative “bad”. But, if these activities are complements, then a further cost of smuggling is that it increases not only cigarette consumption but also alcohol consumption as well. We find some evidence using legal sales data that cigarettes and alcohol are complements; higher cigarette taxes lead to lower alcohol consumption, and that smuggling lead to increased alcohol consumption. We do not find confirmatory evidence in the expenditure data, however; thus, while it is clear that these goods are not substitutes, there is mixed evidence on whether they are complements.
Our paper proceeds as follows. Part 2 provides some background on the literature on the price elasticity of demand for cigarettes, and on cigarette taxation in Canada and the US. Part 3 describes our data, and Part 4 discusses our empirical methodology. Part 5 presents our basic results on demand elasticities. Part 6 discusses estimated elasticities by income group, and Part 7 presents evidence on the substitutability of cigarettes and alcohol. Part 8 concludes.
Section snippets
Background
There is an enormous literature in the United States that has been devoted to estimating the price elasticity of demand for cigarettes. This literature has taken advantage of the fact that there is substantial variation in the price of cigarettes across the US states, and significant changes within states over time, due to variable state excise tax policies. For example, in 2001, the excise tax on cigarettes varied from a low of 2.5 cents in Virginia to a high of US$ 1.11 in New York.
The US
Tax, price, and legal sales data
There exists significant cross-province and time-series variation in both federal and provincial cigarette taxes. Federal taxes consist of excise taxes, excise duties, and sales taxes (the federal sales tax until 1991, and the goods and services tax (GST), thereafter). There is considerable variation in these taxes over time. Moreover, unlike other federal taxes, federal taxes on cigarettes also differ across provinces at a point in time. Specifically, while federal excise duties and sales
Empirical methodology
Our basic empirical methodology follows that used in the previous literature. We use the legal sales data to estimate models of the form:where SALES is cigarettes sold per capita in province j in year t; PRICE is the cigarette price; δj and τt are fixed effects for province and year, respectively; and TIME is a linear time trend. By including province fixed effects, we control for any fixed differences in tastes for smoking across provinces. Likewise, by
Basic results
Our basic estimates from the legal sales data, with no attempt to correct for smuggling, are shown in the first column of Table 3, which shows a linear model which relates cigarette sales per capita to the price (instrumented by tax). We estimate that for every CAN$ 1 rise in (real) price, there is a 49.95 reduction in cigarette consumption per capita. The implied elasticity is −0.72. This estimate is somewhat larger than the elasticity estimates for the US; given that prices are so much higher
Motivation
A primary consideration in the policy debate over excise taxation is the potential regressivity of tobacco taxes. Lower income groups spend a much larger share of their incomes on cigarettes than do higher income groups. This is illustrated for Canada in Table 5. The first column of this table shows the distribution of cigarette expenditures as a share of after-tax income by after-tax income quartile in 1998. While the lowest income quartile spent 4% of after-tax income on cigarettes, the
Effect on alcohol consumption
Another key issue that has not been explored in the Canadian context, and has also been relatively unexplored in the US context as well, is the substitutability or complementarity of cigarette and alcohol consumption. Ex ante, it is not clear if these activities will be complementary or substitutable. If there is a fixed demand for such “vice” activities, then when cigarette prices go up, individuals will substitute into drinking. But, if the activities are pursued together, or if individuals
Conclusion
A central parameter for designing public policy towards smoking is the price elasticity of demand for cigarettes. While there are numerous credible attempts to estimate this parameter in the US, there is much less evidence for Canada. Partly this is because the key period of price variation in Canada is a period of significant smuggling.
We have presented two approaches to surmounting the bias to estimated elasticities from smuggling. The first is to use legal sales data, and exclude the regions
Acknowledgements
We are grateful to Mark Bergman for excellent research assistance. Sen gratefully acknowledges funding support from the Centre for Behavioural Research and Program Evaluation of the Canadian Cancer Society/National Cancer Institute of Canada. Stabile gratefully acknowledges support from CIHR (Grant #MOP-53133).
References (20)
The complementarity of teen smoking and drinking
Journal of Health Economics
(1999)- et al.
Taxation smuggling and demand for cigarettes in Canada: evidence from time-series data
Journal of Health Economics
(1997) - et al.
A theory of rational addiction
Journal of Political Economy
(1988) - et al.
An empirical analysis of cigarette addiction
American Economic Review
(1994) - Criminal Intelligence Service Canada, 1999. Annual Report, 1999. Available at...
- Chaloupka, F., Warner, K.E., 2000. The economics of smoking. In: Newhouse, J.P., Cuyler, A.J. (Eds.), The Handbook of...
- Evans, W., Ringel, J., Stech, D., 1999. Tobacco taxes and public policy to discourage smoking. In: Poterba, J. (Ed.),...
- Gruber, J., Koszegi, B., 2000. Is Addiction Rational? Theory and Evidence. NBER Working Paper...
- et al.
Is addiction rational? Theory and evidence
Quarterly Journal of Economics
(2001) - Gruber, J., Koszegi, B., 2002. A Theory of Government Regulation of Addictive Bads: Optimal Tax Levels and Tax...
Cited by (120)
Resolving the milk addiction paradox
2021, Journal of Health EconomicsThe smoking epidemic across generations, genders, and educational groups: A matter of diffusion of innovations
2019, Economics and Human BiologyTobacco excise tax reform: From the perspective of “tax to control tobacco” policy in China
2023, Health Economics (United Kingdom)Social determinants of tobacco smoking among male population in an emerging economy
2023, International Journal of Environment and HealthThe Economics of Tobacco Regulation: A Comprehensive Review†
2022, Journal of Economic Literature