Distribution

The analysis of the overall US SSB market for 2018 revealed a number of ‘spikes’ in the sales volume distribution of beverages by sugar content per 8-oz serving. Panel A in Fig. 1 shows that by far the largest single portion of SSBs sold were for those with 26 g of sugar/8-oz serving (almost 4 billion gallons; 11.8 gallons per capita), such as sodas and energy drinks. A distant second and third at just over 1.5 billion gallons each were beverages with 28 g and 27 g of sugar/8-oz serving, again made up mainly from sodas and energy drinks. Overall, 70.8% of SSB sales were for products in the high-sugar range of ≥ 25 g of sugar/8-oz, with 66.9% of SSB sales in range of 26–31 g of sugar/8-oz. Next, there were two clusters of sales volume by sugar content observed in the middle part of the distribution: one cluster with a peak at 14 g of sugar/8-oz serving (16.9% in the 10–15 g range) that included mostly sports drinks and value-added (enhanced) water, fruit drinks, and RTD teas/coffees. Another smaller cluster occurred around 19 g of sugar/8-oz (8.7% in the 16–20 g range) including mostly fruit drinks and RTD teas/coffees. Volumes for calorically sweetened beverages with < 10 g of sugar/8-oz were very low (1.1%).

Fig. 1

Distribution of annual sugar-sweetened beverage (SSB) sales volume by sugar content, all SSBs and by SSB category, US total, 2018

There were considerable differences in the volume distribution by sugar content across beverage categories (shown in Panels B through F in Fig. 1). Almost the entire volume of sodas (98.1%) and energy drinks (97.3%) was for high-sugar products with ≥ 25 g of sugar/8-oz. For fruit drinks, 60.6% of sales were in the range of 12–20 g of sugar/8-oz and 27.3% in the high-sugar range. The volume of RTD teas/coffees spanned across the sugar content spectrum and sports drinks and value-added waters was mostly clustered at 12–14 g of sugar/8-oz (91.3%).

SSB tiered tax structure

The sugar content analysis revealed a dominant cluster of beverage sales in the high range of ≥ 25 g of sugar/8-oz. Next, we observed a cluster in the mid-range of 10–15 g, with an adjacent cluster from 16 to 20 g of sugar/8-oz. The identification of these clusters across the sugar content spectrum constitutes a population data-based approach for the development of thresholds for sugar content in a tiered tax structure for SSB taxes. To discourage consumption of high-sugar beverages and incentivize industry reformulation, sugar content thresholds for differential tax amounts per unit of volume should be implemented at meaningful distances below the major cluster ranges. In this regard, a public health action plan that was foundational to the development of the SSB tiered tax in the UK set a goal to reduce sugar levels in foods and beverages by 20% [20]. This would suggest a reduction in sugar for the cluster of high-range (≥ 25 g of sugar/8-oz) sugar content beverages by a minimum of 5 g to a threshold of under 20 g of sugar/8-oz. Using this same level of reduction suggests a threshold of under 5 g of sugar/8-oz for the lower bracket. This would place it at equal distance from the 10 g lower bound of the mid-range cluster, although there is a larger percentage change in relation to the mid-range cluster. Using the 5 g minimum reduction as a basis for a meaningful reduction in sugar content, the pattern of the SSB volume distribution shown in Panel A in Fig. 1 reveals that thresholds for sugar content at 20 g per 8-oz (equivalent to 8.4 g per 100 ml) and 5 g per 8-oz (equivalent to 2.1 g per 100 ml) may reduce consumption and incentivize reformulation. Thus, based on the observed volume distribution by sugar content, we recommend the following three-tiered tax structure:

Tier 1 (T1):

No tax for SSBs with sugar content of > 0 g and < 5 g per 8-oz

Tier 2 (T2):

Lower tax rate for SSBs with sugar content of ≥ 5 g and < 20 g per 8-oz

Tier 3 (T3):

Higher tax rate for SSBs with sugar content of ≥ 20 g per 8-oz

Based on the distribution of SSB volume by sugar content, only 0.1% of volume sold currently would fall into T1 and not be taxed, 25.3% would fall into T2, and 74.6% would fall into the highest tax tier of T3. For the overwhelming majority of SSBs, that contain high amounts of sugar at ≥ 25 g of sugar/8-oz, reformulation that reduces added sugar in these beverages to under 20 g/8-oz would represent a 20% to 38% reduction in sugar content.

Panels A through D in Fig. 2 depict the SSB volume distribution across the tax tiers by gender, age, race/ethnicity, and education. Panel A shows that males account for more of the volume in both T2 and T3 compared to females, but similarly for both genders, roughly three times the SSB volume falls into T3 versus T2. Panel B shows considerable variation across the tiers by age. Young adults (20–44-year-olds) consume the highest per capita volume of T3 beverages whereas youths aged 10–19 consume the highest per capita volume of T2 SSBs, including a substantial portion of sports drinks. Further, 2.3 to 3.5 times the SSB volume per capita falls into T3 versus T2 for all age groups, except children aged 0–9 who have almost similar volumes in T3 and T2. By race/ethnicity, Panel C shows that non-Hispanic black and Hispanic persons have the highest absolute per capita volumes in T3, and Hispanics have the highest relative consumption of 3.8 times the volume in T3 versus T2. Non-Hispanic black persons also consume the highest absolute volume of beverages in T2. By education (for adults only), Panel D shows that lower-educated adults with no college consume the highest per capita volume of SSBs in T3 while also having the highest proportion of volume in T3 versus T2; a ratio of 3.9 and 3.5 for those with less than high school-level education and high school diploma, respectively, compared to 2.4 and 2.2 times for their counterparts with some college and college degree or more, respectively.

Fig. 2

Per capita 2018 sales volume of sugar-sweetened beverages by tax tiers, by gender (a), age group (b), race/ethnicity (c), and education (d)

Tax revenue

Table 1 shows that increasingly higher flat tax amounts per SSB ounce of 1¢/oz, 1.5¢/oz, and 2¢/oz are estimated to raise $13.9 billion, $18.0 billion, and $20.1 billion dollars in tax revenue, respectively. Note that as the flat tax rate doubles, for example, from 1¢/oz to 2¢/oz, tax revenue less than doubles due to the predicted decline in consumer demand that is assumed to be elastic. A flat tax of 2¢/oz versus 1¢/oz raises 44% more tax revenue and results in lower net SSB volume (7.9 versus 10.9 billion gallons, respectively; data not shown). Table 1 also provides flat tax and tiered tax revenue estimates by state.

Table 1 Sugar-sweetened beverage (SSB) tax revenue for the USA and by state, by flat tax and tiered tax structures, 2018

A tiered tax of 1¢/oz for T2 SSBs with sugar content of ≥ 5 g and < 20 g per 8-oz and 2¢/oz for T3 SSBs with sugar content of ≥ 20 g per 8-oz is projected to raise $18.2 billion in total SSB tax revenue across the two tiers, which is just slightly higher than the tax revenue projection of $18.0 billion for a flat tax amount of 1.5¢/oz (the mean of the two tiered amounts) on all SSBs. The tiered tax is projected to lead to 9% lower SSB volume as compared to the 1.5¢/oz flat tax (8.5 versus 9.4 billion gallons, respectively) due to a higher tax on a greater proportion of the overall volume in T3 and a related larger reduction in SSB volume.

In sensitivity analyses, we re-estimated tax revenue for the tiered tax to account for potential reformulation in each beverage category, keeping all other model assumptions the same. Given that reformulation led to an 11% reduction in sugar content in beverages following the introduction of the SDIL in the UK, we first considered reformulation where 10% of volume in T3 moves into T2 and 10% of volume in T2 moves into T1. Next, because the UK’s top tier tax rate is slightly less than one half of the 2-cent/oz top tier rate in this study, we considered a second scenario where 20% of volume in T3 moves into T2 and 20% of volume in T2 moves into T1. Overall, tax revenue from the tired tax fell from $18.2 billion with no reformulation to $17.4 billion, a 4.3% reduction in revenue, under the assumption of 10% reformulation, to $16.6 billion, a 8.9% reduction in revenue, under the assumption of 20% reformulation (data not shown in tables). As expected, under both reformulation scenarios, volume was lower in T3 and higher in T2 and T1. It is higher in T2 because the shift in volume from T3 to T2 was greater than the shift from T2 to T1 (since the greatest volume originated in T3). Thus, although revenue is lower with reformulation, so too is the volume of beverages falling into the highest sugar content tier: under the tiered tax (compared to no tax) beverage volume in T3 falls by 45.4% assuming no reformulation, by 50.9% under the 10% reformulation assumption, and by 56.4% under the 20% reformulation assumption.