No smoking gun: tobacco taxation and smuggling in Sierra Leone

Objective To evaluate the common industry claim that higher tobacco taxation leads to higher levels of smuggling, particularly in a limited state capacity setting. Design This paper evaluates the effects of a tobacco tax increase in Sierra Leone on smuggling by using gap analyses. Its models are based on multiple rounds of the Demographic and Health Survey and customs data as well as newly collected data on cigarette prices. Results The paper shows that despite a substantial increase in cigarette taxation, and despite the absence of other formal tobacco control policies, smuggling has not increased in Sierra Leone. Its primary model shows a decrease in cigarette smuggling by 16.74% following the tax increase, alongside a decrease in cigarette consumption more widely and an increase in tax revenue. Conclusions By presenting a low income and lower enforcement capacity case study, this paper provides novel and critical evidence to the debate on the tax-smuggling link. Furthermore, it points to new questions on how states in these contexts can limit cigarette smuggling.

reported include the various charges that existed at the time.This is an issue, as in order to recover the true quantity of imported cigarettes we need to have pre-tax unitary prices -posttax unitary prices are always higher than pre-tax ones, and this would lead us to recover a quantity of imported cigarettes lower than the actual one.As we cannot be sure, we calculate the gap for 2013 both assuming that the reported prices are pre-tax and assuming the they are post-tax, with latter always leading to higher imported volumes and, everything else equals, lower gaps.This is not an issue in the 2021 wave of the survey, as it was explicitly asked to respondents to report prices net of taxes if they were importers.
2. Construction of average cigarette price.Both rounds of the price survey include information about brand demand, as well as including several cigarettes importers.From this information, we construct three different price measures.One is a simple average of all cigarette prices BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s) from wholesaler, one is a weighted average of wholesalers' prices, where high-demand branded are weighted at 50%, medium-demand brands at 30% and low demand brands at 20%.Finally, we also construct a simple average price using only information from importers.Simple average prices are higher than weighted average ones, with importer-only averages being the lowest.As above, the higher the price, the lower the quantity of imported cigarettes, the higher the gapeverything else equal.
3. Treatment of non-daily smokers.While the DHS 2013 did not contain information about smoking frequency, the 2019 round does, and information about smoking intensity is only widely available for daily smokers.This poses issues of comparability, so that non-daily smokers are always included in the estimates, although in two different ways.In the "floor" scenario, non-daily smokers only smoke one cigarette every other day, while in the "ceiling" scenario they consume half as many cigarettes as a daily smoker.That is, we apply the same distribution of daily cigarette consumption of daily smokers to non-daily ones, but with smoking levels reduced to a half.In the "floor" scenario, the number of cigarettes consumed is lower, so that, everything else equal, the gaps are also lower.

Inclusion of senior citizens. The Demographic and Health Survey (DHS) only covers male
respondents aged between 15 to 59 and female respondents aged 15 to 49, as it mostly targets information about reproductive health.However, there are no reason to assume that senior citizens do not smoke, so that they should reasonably be included in the estimates.In order to do so, we obtained the distribution of the overall population by gender and age cohort from the Housing and Census Population 2015 and from the Sierra Leone Integrated Housing Survey 2018, both implemented by Statistics Sierra Leone.Both reports included population growth rates, which we assumed to be homogenous across age cohorts.With this assumption, we extrapolated the dimension of the missing population brackets, using the 2015 census to augment the DHS 2013 and the 2018 survey to augment the DHS 2019.After obtaining the gendered distribution of smoking intensities from the two DHS, we applied it to the newly obtained populations.That is, we assume that both smoking incidence and the distribution of smoking intensity for the male cohort of those aged 60-80+ and the female cohort of those BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s) aged 50-80+ are the same of those of the male cohort aged 15-59 and the female cohort aged 15-49 respectively.As a consequence, everything else equal, excluding senior citizens lead to lower cigarette consumption, hence to lower gaps.5. Under-reporting of cigarettes consumed.It is widely thought that self-reported levels of cigarette consumption are under-estimated by respondents, due to both recall bias and undesirability of the habit.We consequently proceed to inflate the reported level of consumption by 5% in certain scenarios and 20% in others.Inflating reported consumption increases the number of cigarettes smoked, hence increasing the size of the gap.
Given the above assumptions, we construct 4 sets of scenarios: a) Exclusion of senior citizens, no inflation of reported consumption.b) Inclusion of senior citizens, no inflation of reported consumption.c) Exclusion of senior citizens, inflation of reported consumption by 5% d) Inclusion of senior citizens, inflation of reported consumption by 20%.
In each of the above sets of scenarios, we then calculate the difference between the gap in 2013 and the gap in 2019 using all constructed average prices -simple average, weighted average, importers-