Starting from the first part of the reform of immediate application (year 1), we have increased the rate of excise duty so that the total tax (excluding VAT) is equal to 75% of the TIRSP. As the excise rate increases, the import duty rate is phased out. The average excise duty per pack increased by 88%, resulting in a 44% increase in the average price (Table 3).

Cigarette sales and industry revenue are expected to drop 11% and 10% respectively. The change in tax revenues is expected to be even more pronounced. Specifically, excise and VAT revenues are expected to increase by 67% and 28%, respectively. Revenue from import duties, however, as expected, will be reduced by 11%. Total tax revenue will increase by 44%. The number of cigarette smokers is expected to decrease by 5% and the overall smoking prevalence will decrease by 1%.

Due to the ad valorem nature of the excise tax, changes in key market indicators are expected to be more pronounced for premium brands and weaker for economy brands. The price dispersion index is 38% since the most expensive brand is estimated at SAR 47.37 and the cheapest brand at SAR 18.05. The price dispersion index is relatively low, which creates downside trading opportunities.

On average, excise revenue increases by 67%, but the corresponding increase by price segment is 70% for premium brands, 62% for mid-priced and 54% for economy brands. After the tax reform, the total tax is around 80% of the TIRSP on average, with this share being higher for low-priced cigarettes (around 87%) due to the minimum import duty to which they are subject.

The next step (year 2, in Table 3) is to replace import duties with excise duties keeping the total tax share (excluding VAT) constant, i.e. setting the excise duty at 75% of the TIRSP and to introduce a MET at 70% of the weighted average price. (WAP). The introduction of MET (SAR 28) is impacting mid-priced and economy brands. The TIRSP percentage excise tax is 76% for mid-priced brands and 87% for economy brands. Thus, the MET has a significant effect, especially on economy brands.

This reform would result in a further 8% increase in the average price, a 4% reduction in sales, a 4% increase in total cigarette tax revenue and, more specifically, a 27% increase in excise revenue and a 3% increase in VAT receipts. In addition, it will result in a 2% reduction in the number of smokers with a 0.4% reduction in smoking prevalence.

Finally (year 3, in Table 3), a move to a revenue-neutral specific tax rate is discussed. The ad valorem rate is replaced by a specific excise duty so that excise duty revenue remains constant. According to our simulations, this corresponds to a specific excise at SAR 31. Even when we pass a tax reform that keeps excise revenue constant, the change in tax structure is expected to lead to an additional 1% increase in the average package price, 1 % reduction in sales and 0.3% reduction in the number of smokers with a 0.1% reduction in prevalence. Establishing a higher specific rate will lead to further reductions in sales and increased tax revenues. If one sets a higher specific excise duty (than that which guarantees constant tax revenue), for example SAR 35, it is estimated that sales fall by 3% and total revenue increases by 10%.

In Table 3, we report only the (sales-weighted) change in price on average (1% increase). However, to have some intuition, we need to look at what is happening in the three price segments individually. The revenue-neutral replacement of the ad valorem tax by the specific tax has distinct effects on the three price segments. The average price of the premium segment fell (−3%) while the average price of the medium and economy segments increased (9 and 10% respectively). As expected, the drop in sales came from mid-range and economy brands. Due to the tax change, the excise tax share of these segments increases, as does their contribution to tax revenue. On average, revenue is constant, as the increase in revenue from the mid-priced and economy segments is compensated by the loss of revenue from the premium segment (due to the decreasing share of excise duties and their inelastic demand) .

Overall, the three-year reform would lead to a more than 50% increase in cigarette prices, a 16% reduction in cigarette sales, and an almost 50% increase in total cigarette tax revenue. The final total share of taxes would be 81% and the share of excise duties 76% of the final price (all taxes included).

Using estimate that adult smokers were 2,676,978 in 2017 [20], and assuming an overall price elasticity of demand equal to −0.3, we also estimated the number of deaths averted. Based on the standard estimate that the elasticity of smoking prevalence is half of the total elasticity of demand, that one in two regular smokers will eventually die and all smokers will survive [21], we estimated that 88,340 tobacco-related deaths would ultimately be averted through the first year of tax reform. In other words, assuming a prevalence elasticity of −0.15, a tax increase that would lead to a 44% increase in price would ultimately lead to a 6.6% reduction in cigarette-related deaths. Assuming a higher elasticity of demand, of course, this would lead to more deaths averted. For example, for a total elasticity of demand equal to −0.4, cigarette-related deaths would decrease by 8.8% (117,787 deaths would ultimately be avoided).

### Sensitivity analysis

Our elasticity assumptions are rather conservative. By increasing the elasticity of demand for cigarettes by price segment, the estimated reduction in smoking prevalence is higher. Assuming, for example, a demand elasticity of −0.3, −0.4, and −0.5 for premium, mid-price, and economy brands respectively (scenario 1), smoking prevalence would decrease by 1.4 % (− 2.3% over the 3-year period). Assuming a demand elasticity of −0.4, −0.5, and −0.6 for premium, mid-price, and economy brands respectively (scenario 2), smoking prevalence would decrease by 1.8% (−2 .5% over the 3-year period).

Obviously, according to the elasticity hypotheses, there is a trade-off between a larger drop in sales and therefore in the number of smokers and in the prevalence rate, and a smaller increase in tax revenue. In Scenario 1 and over the 3-year period, sales would drop by 21%, and excise tax revenue and total tax revenue would increase by 84% and 40%, respectively. In Scenario 2 and over the 3-year period, sales would drop by 28%, and excise tax revenue and total tax revenue would increase by 75 and 32%, respectively.

Finally, assuming no downside trade, we overestimate the reduction in sales and therefore underestimate the increase in tax revenue. When we allow some price decline, that is, consumers switching to cheaper brands as prices rise, our results do not change significantly. In the absence of solid data, it is safer not to make arbitrary assumptions about declines or rises.