Targeted Beverage Taxes Influence Food and Beverage Purchases among Households with Preschool Children^,,

Abstract

Background: How beverage taxes might influence purchases of foods and beverages among households with preschool children is unclear. Thus, we examined the relation between beverage taxes and food and beverage purchases among US households with a child 2–5 y of age.

Objectives: We examined how a potential tax on sugar-sweetened beverages (SSBs), or SSBs and >1% fat and/or high-sugar milk, would influence household food and beverage purchases among US households with a preschool child. We aimed to identify the lowest tax rate associated with meaningful changes in purchases.

Methods: We used household food and beverage purchase data from households with a single child who participated in the 2009–2012 Nielsen Homescan Panel. A 2-part, multilevel panel model was used to examine the relation between beverage prices and food and beverage purchases. Logistic regression was used in the first part of the model to estimate the probability of a food/beverage being purchased, whereas the second part of the model used log-linear regression to estimate predicted changes in purchases among reporting households. Estimates from both parts were combined, and bootstrapping was performed to obtain corrected SEs. In separate models, prices of SSBs, or SSBs and >1% and/or high-sugar milk, were perturbed by +10%, +15%, and +20%. Predicted changes in food and beverage purchases were compared across models.

Results: Price increases of 10%, 15%, and 20% on SSBs were associated with fewer purchases of juice drinks, whereas price increases of 10%, 15%, and 20% simulated on both SSBs plus >1% fat and/or high-sugar milk (combined tax) were associated with fewer kilocalories purchased from >1% fat, low-sugar milk, and meat, poultry, fish, and mixed meat dishes.

Conclusions: Our study provides further evidence that a tax on beverages high in sugar and/or fat may be associated with favorable changes in beverage purchases among US households with a preschool child.

Introduction

Childhood obesity is a leading threat to public health in the United States (1). Because treatment of childhood obesity remains challenging (2, 3), prevention is instrumental in reducing the overall burden of this public health problem in the United States (4–7). Preschool children (ages 2–5 y) are an ideal population for dietary interventions because eating behaviors and food preferences are formed during the first 5 y of life (8, 9), and because the food environments of young children are controlled by parents and/or adult caregivers. There may also be greater opportunity to prevent excess weight gain in preschoolers, among whom the prevalence of obesity is lower than that of older children (1). US preschooler diets are high in beverages like high-fat (>1% fat) milk and sugar-sweetened beverages (SSBs) (10). These beverages, in particular, represent a key concern because they are thought to contribute to excess total energy intake by having a smaller relative effect on satiety than food (11–16). Thus, consuming too many kilocalories from beverages over time can lead to weight gain (17). Taxing beverages has been an option proposed by a number of childhood obesity researchers and health advocates as a means to limit the consumption of beverages high in fats and/or sugars (18–24). However, there is minimal research on whether such taxes might influence food and beverage purchases among households with preschool-aged children.

Taxes on certain beverages may have unintended consequences such as increasing purchases of other beverages or foods high in fats/sugars, and/or decreasing purchases of healthier beverages such as low-fat, low-sugar milk and 100% fruit juice (25, 26). Previous studies focus primarily on SSBs, comprising caloric soft drinks, juice drinks, and sport and energy drinks (25–30). While an SSB tax has been associated with fewer SSB purchases (25–29), only 2 of these studies also considered the relation between an SSB tax and purchases of foods. To understand the overall implications of an SSB tax, including whether such a tax could influence net weight of foods/beverages purchased, the relation between beverage taxes and purchases of both beverages and foods must be examined. It had also been suggested that higher-fat (>1% fat by weight) and/or high-sugar milk be taxed (23, 24) because intake of these beverages is discouraged for children >2 y of age (31). Nonetheless, we were unaware of any prior study in which taxes on >1% fat and/or high-sugar milk had been explored, thus, it is unclear if taxing these beverages would affect purchases of these and other foods/beverages.

To address these important gaps in the literature, we used data from the 2009–2012 Nielsen Homescan Panel to simulate “taxes” by increasing the prices of selected beverages by 10%, 15%, and 20%. We began by examining the association between simulated price increases on SSBs and purchases of an array of consumer-packaged foods and beverage groups among US households with a preschool child. Next, we compared these associations with those observed when simulating price increases on SSBs plus >1% fat and/or high-sugar milk. We then extrapolated our findings to estimate the associations between each beverage tax and annual weight of foods/beverages purchased per capita. By addressing these key gaps in the literature, we aimed to provide further evidence to inform policy decisions regarding the use of targeted beverage taxes as a potential means to reduce purchases of unhealthful beverages among US households with preschool-aged children.

Methods

Sample.

We included quarterly household purchase data from households in 76 metropolitan and nonmetropolitan areas (markets) from the 2009–2012 Nielsen Homescan Panel (The Nielsen Co.). Homescan comprises a representative panel of US households who report weekly consumer-packaged goods purchases with use of barcode scanners issued by Nielsen. Purchases without barcodes (e.g., fresh fruits and vegetables) were not included in this analysis. Further details regarding the sample have been published elsewhere (32–35).

To minimize heterogeneity in our findings because of household composition, we limited our analyses to households with 1 child between the ages of 2 and 5 y. Data were included for these households who participated in Homescan during at least 1 quarter between quarter 1 of 2009 and quarter 4 of 2012. We used a threshold of >5 SDs from the mean weight total consumer-packaged foods and beverages purchased per capita to exclude outliers (n = 56 quarterly observations) (30). Our final sample included 57,283 quarterly observations from 14,784 unique households (Table 1).

TABLE 1.

Sample characteristics of households with a preschool child from Nielsen Homescan Panel, 2009–2012¹

	2009	2010	2011	2012
Total observations, n	15,088	14,272	13,756	14,167
Unique households, n	3892	3689	3557	3646
Total per capita purchases, g/d
Sugar-sweetened beverages	131 ± 6.4	127 ± 3.1	108 ± 1.7*	106 ± 1.9*
All beverages	501 ± 4.0	480 ± 4.1*	459 ± 4.0*	447 ± 3.7*
All foods	344 ± 2.1	341 ± 2.2	330 ± 2.2*	330 ± 2.1*
Total per capita purchases, kcal/d
Sugar-sweetened beverages	88.7 ± 3.4	86.9 ± 2.4	74.2 ± 1.4*	71.0 ± 1.6*
All beverages	140 ± 1.2	132 ± 1.2*	121 ± 1.1*	120 ± 1.1*
All foods	777 ± 4.9	763 ± 5.1	735 ± 4.8*	740 ± 4.7*
Race/ethnicity, %
Non-Hispanic white	67.4 ± 0.6	66.8 ± 0.6	67.9 ± 0.6	66.8 ± 0.5
Non-Hispanic black	9.1 ± 0.3	9.5 ± 0.4	9.9 ± 0.4	10.3 ± 0.3*
Hispanic	16.9 ± 0.5	17.0 ± 0.5	15.3 ± 0.5	16.2 ± 0.4
Head of household education, %
<High school	1.2 ± 0.1	1.4 ± 0.2	1.8 ± 0.2*	1.4 ± 0.1
High school	16.8 ± 0.4	15.0 ± 0.4*	15.7 ± 0.5	13.3 ± 0.4*
Some college	30.9 ± 0.5	30.7 ± 0.6	29.4 ± 0.6	31.2 ± 0.5
College graduate	51.1 ± 0.6	53.0 ± 0.6	53.0 ± 0.6*	54.0 ± 0.6*
Household income, %
≤100% FPL	10.1 ± 0.4	11.0 ± 0.4	10.9 ± 0.4	12.1 ± 0.4*
>100–130% FPL	7.6 ± 0.3	8.0 ± 0.3	8.5 ± 0.3	7.3 ± 0.3
>130–185% FPL	13.4 ± 0.4	11.7 ± 0.4*	13.1 ± 0.4	16.8 ± 0.4*
>185−400% FPL	51.0 ± 0.6	50.3 ± 0.6	48.8 ± 0.6*	45.7 ± 0.5*
>400% FPL	17.8 ± 0.4	19.0 ± 0.5	18.7 ± 0.5	18.2 ± 0.4

¹Values are means ± SEs unless otherwise indicated. *Different from 2009, P < 0.05. FPL, federal poverty level. University of North Carolina calculation based in part on data reported by Nielsen through its Homescan Services for all food categories, including beverages and alcohol for the 2009–2012 periods, for the US market. Copyright © 2015 The Nielsen Company.

Food and beverage groups.

The Nielsen Co. categorizes foods and beverages into “modules” comprising foods/beverages with similar commercial properties. As the main focus of this work, we classified beverages with the overarching goal of creating beverage groups of public health relevance. A further goal was to disaggregate broad beverage groups commonly used in the literature (e.g., SSBs) in an effort to better reflect the heterogeneity within beverage groups. Therefore, 10 mutually exclusive beverage groups were created with use of Homescan “modules,” product ingredient lists, product claims, and Nutrition Facts Panel information (36): 1) caloric soft drinks; 2) sport and energy drinks; 3) juice drinks; 4) >1% fat, high-sugar milk; 5) >1% fat, low-sugar milk; 6) low-fat, high-sugar milk; 7) low-fat, low-sugar milk; 8) 100% juice; 9) diet drinks; and 10) tap, bottled, and flavored waters. The 2010 US Dietary Guidelines for Americans (31), the American Academy of Pediatrics (37), the Institute of Medicine (23), and the 2015 US Dietary Guidelines Advisory Council (38) all advise that intakes of >1% fat milk and milk containing added sugar be limited for children ages 2 and older. Moreover, the Institute of Medicine defines high-sugar beverages as those containing ≥22-g sugar per 8-oz (236.59 mL) serving (23). Milk subgroups by fat and sugar content were therefore based on these guidelines. Additionally, summary groups were created for total beverages, total foods, total SSBs—which include caloric soft drinks (excluding low-calorie soft drinks), fruit/juice drinks (not containing 100% juice), and sport and energy drinks (excluding low-calorie options) —and total non-SSBs. A detailed description of our approach to classifying beverages is shown in Supplemental Table 1. All dried and concentrated beverages were reconstituted to “ready-to-eat” form with use of standard conversion factors (29.35 g per ounce for dry weights, and 29.57 mL per ounce for liquids) and manufacturer’s reconstitution instructions (e.g., frozen concentrated juice). A total of 9 comprehensive, mutually exclusive food groups were created with use of Homescan “modules”: 1) dairy products excluding milk; 2) meat, poultry, fish, and mixed meat dishes; 3) other proteins; 4) grain products (excluding ready-to-eat desserts); 5) fruits and vegetables; 6) fats, oils, sauces, and condiments; 7) sweets and snacks; 8) “other” foods; and 9) mixed dishes and soups.

Prices.

Market-quarterly food/beverage prices per 100 g were derived with use of purchase data and prices paid. Additionally, to control for differences in the cost of living by market and quarter, a Food Price Index was created with use of quarter 1 of 2000 in Los Angeles, California, as the referent index. A detailed description of these methods is given in Supplemental Table 2. Prices for selected beverages are also shown in Supplemental Table 3.

Unemployment rate.

Market-quarterly unemployment rates were used to reflect the economic conditions for participating Homescan households (39). Using data from the Bureau of Labor Statistics’s local area unemployment statistics (40), quarterly unemployment rates were computed by taking the mean unemployment rate for the 3 mo comprising each quarter (from quarter 1 of 2009 to quarter 4 of 2012) for the 76 markets (41).

Statistical analysis.

All analyses were conducted in STATA (version 13, 2011; StataCorp), and survey weights were used in all calculations to adjust for differential probability of selection. For food and beverage groups purchased by <80% of households, a right-skewed distribution and a preponderance of zeros require special consideration. It has been previously shown that a 2-part marginal effects model is an appropriate statistical approach for dealing with such data (25, 42, 43). Thus, a 2-part marginal effects model, comprising probit, and ordinary least squares regression, was used to estimate the relation between price and amount purchased (42). For food/beverage outcomes reported by ≥80% of included households, only the second part of the model [ordinary least squares (OLS) regression] was used. In the first part of the 2-part model, probit regression was used to model the probability of a household purchasing the outcome food/beverage of interest. In the second part, conditional OLS regression was used to model the amount purchased among households reporting nonzero expenditures. Coefficients from both parts of the model were algebraically combined to estimate the amount purchased associated with simulated taxes on selected beverages among all households with a preschooler. To obtain corrected SEs, models were clustered at the market level, and bootstrapping was performed (1000 replications) to account for correlation resulting from repeated measurements (44) and potential correlation between households in the same market. For food and beverage groups purchased by ≥80% of the sample, only the second part (OLS regression) of the 2-part model was used.

In all models, prices were log-transformed with use of the natural log. In OLS regression models, food and beverage prices and amount purchased per capita from each food/beverage group were log-transformed to simplify model interpretation (log-log model), and in keeping with prior works (26–30, 45). To account for error that may arise when outcome variables are log-transformed (46), we multiplied predicted values (e.g., predicted amount purchased with a 20% increase in SSB price) by the appropriate Duan Smearing estimator upon retransformation with use of the anti-log (47). Elasticities were ascertained from untransformed model coefficients, and thus, Duan smear factors were not applied to these values. In separate multilevel models, price increases of 10%, 15%, and 20% were simulated for the following: 1) SSBs alone and 2) SSBs plus >1% fat and/or high-sugar milk. A conservative threshold for statistical significance (α = 0.10) was used for continuity with related prior studies (25–28, 30, 48) and to account for measurement error (35).

To simulate a “tax” on selected beverages, the prices of these beverages were perturbed in statistical models, assuming 100% transference of the tax to shelf price. “Simulated tax” is used henceforth to refer to increases in the price paid. Separate models were run with each food/beverage group of interest as the outcome, and all models were adjusted for household composition (number of household members by gender and age: 2–5 y; 6–11 y; 12–18 y; and ≥19 y); household income as a percent of federal poverty level (FPL) (≤100% FPL; >100–130% FPL; >130–185% FPL; >185–400% FPL; and >400% FPL), education level (highest level of education completed by head of household), race/ethnicity, market-quarterly unemployment rate, year, and quarter. In addition, we tested (using joint Wald test) and found significant interactions (P < 0.10) between price and year for regular soft drinks; juice drinks; low-fat, high-sugar milk; >1% fat, low-sugar milk; >1% fat, high-sugar milk; and sport and energy drinks. Thus, interaction terms for prices of these beverages and year were included in all models. Lastly, following previous works (26–29), assuming changes in total weight of food/beverages purchased per day per capita would be constant over time, the net effect of each “tax” simulation on total foods/beverages purchased per year was estimated by multiplying adjusted estimates of changes in daily purchases of total weight by 365.25.

Results

Demographic characteristics.

Characteristics of the sample, including kilocalories and grams purchased per capita from SSBs by year, are shown in Table 1. Sample households were predominantly non-Hispanic white, with college-educated heads of household, and a household income of >185–400% FPL. Total SSB purchases, total beverage purchases, and total food purchases decreased over time (Bonferroni adjusted, P < 0.05).

Survey-weighted mean amounts of each beverage purchased per capita and amount purchased among reporting households are shown in Figure 1A, B. Households with a preschool child purchased fewer total grams of beverages in 2012 than in 2009. Mean prices by market and percent of household reporting purchases of each beverage are shown in Supplemental Table 3. More than 80% of sampled households reported purchasing >1% fat, low-sugar milk, and juice drinks, whereas fewer than 80% reported purchasing low-fat, low-sugar milk; low-fat, high-sugar milk; >1% fat, high-sugar milk; 100% juice; soft drinks; bottled and flavored water; sport and energy drinks; and diet beverages.

FIGURE 1.

Mean grams purchased per capita per day among households with a preschool child participating in the Nielsen Homescan Panel, 2009–2012. (A) Mean grams purchased per capita among all households in the sample. (B) Mean grams purchased per capita among reporting households (denominator is reporting households only). All values are mean grams purchased per capita per day. University of North Carolina calculation based in part on data reported by Nielsen through its Homescan Services for all food categories, including beverages and alcohol for the 2009–2012 periods, for the US market. SSB, sugar-sweetened beverage. Copyright © 2015 The Nielsen Company.

Elasticities.

Own-price elasticities, defined here as the change in per capita purchases in grams of a given food/beverage divided by the change in price for the same food/beverage, are presented in Table 2. There were moderate and significant (P < 0.05) own-price relations for juice drinks (−1.01), total >1% fat and/or high-sugar milk (−0.71), >1% fat, low-sugar milk (−0.65), low-fat, low-sugar milk (−0.79), 100% juice (−0.99), and diet beverages (−0.62). These values suggest that purchases of these beverages decrease when their price is increased.

TABLE 2.

Elasticities of demand with respect to the price of beverages (n = 56,963)¹

Elasticity of the quantity, g	Total SSBs	Soft drinks	Sport and energy drinks	Juice drinks	Total >1% fat and/or high-sugar milk	>1% fat, high-sugar milk	>1% fat, low-sugar milk	Low-fat, high-sugar milk	Low-fat, low-sugar milk	100% Juice	Diet beverages	Bottled and flavored water
Total SSBs	−0.29 ± 0.28	0.02 ± 0.16	−0.04 ± 0.03	−0.27 ± 0.26	−0.09 ± 0.25	−0.12 ± 0.12	0.16 ± 0.23	−0.13 ± 0.05***	−0.13 ± 0.17	−0.03 ± 0.19	−0.14 ± 0.17	0.00 ± 0.01
Soft drinks	−0.75 ± 0.51***	−0.15 ± 0.21	0.01 ± 0.05	−0.60 ± 0.47	−0.01 ± 0.46	−0.19 ± 0.17	0.35 ± 0.46	−0.17 ± 0.08**	−0.33 ± 0.34	0.10 ± 0.33	−0.23 ± 0.29	−0.02 ± 0.02
Sport and energy drinks	0.56 ± 0.34*	0.16 ± 0.21	−0.04 ± 0.05	0.44 ± 0.32	0.37 ± 0.33	−0.02 ± 0.14	0.43 ± 0.35	−0.05 ± 0.07	0.15 ± 0.28	0.41 ± 0.32	0.26 ± 0.28	−0.01 ± 0.01
Juice drinks	−1.01 ± 0.39***	0.04 ± 0.24	−0.04 ± 0.04	−1.01 ± 0.35***	0.21 ± 0.30	0.15 ± 0.14	0.17 ± 0.30	−0.12 ± 0.06**	−0.22 ± 0.21	−0.41 ± 0.24*	−0.19 ± 0.23	0.02 ± 0.01*
Total >1% fat and/or high-sugar milk	0.16 ± 0.39	0.12 ± 0.16	0.02 ± 0.03	0.02 ± 0.35	−0.71 ± 0.33**	0.03 ± 0.11	−0.71 ± 0.30**	−0.03 ± 0.06	0.18 ± 0.22	0.49 ± 0.20**	−0.10 ± 0.19	−0.01 ± 0.01
>1% fat, high-sugar milk	−0.23 ± 0.34	0.12 ± 0.19	0.00 ± 0.03	−0.35 ± 0.29	−0.72 ± 0.29***	−0.32 ± 0.10***	−0.32 ± 0.31	−0.09 ± 0.05	0.07 ± 0.23	0.16 ± 0.22	0.08 ± 0.20	−0.01 ± 0.01
>1% fat, low-sugar milk	0.12 ± 0.38	0.07 ± 0.15	0.02 ± 0.03	0.03 ± 0.35	−0.58 ± 0.33*	0.08 ± 0.12	−0.65 ± 0.29**	−0.02 ± 0.06	0.09 ± 0.22	0.33 ± 0.20*	0.02 ± 0.18	0.00 ± 0.01
Low-fat, high-sugar milk	−0.12 ± 0.19	−0.07 ± 0.10	0.02 ± 0.02	−0.07 ± 0.17	−0.08 ± -0.48	0.03 ± 0.07	−0.01 ± 0.18	−0.09 ± 0.04**	0.25 ± 0.14*	0.18 ± 0.13	0.01 ± 0.12	0.01 ± 0.01
Low-fat, low-sugar milk	−1.50 ± 0.58***	−0.18 ± 0.27	−0.05 ± 0.06	−1.28 ± 0.51**	0.54 ± 0.54	−0.11 ± 0.19	0.30 ± 0.55	−0.13 ± 0.08	−0.79 ± 0.38**	−0.79 ± 0.40**	0.24 ± 0.31	−0.06 ± 0.02***
100% Juice	−0.09 ± 0.38	−0.05 ± 0.22	0.00 ± 0.04	−0.04 ± 0.35	0.52 ± 0.33	−0.08 ± 0.13	0.67 ± 0.34*	−0.07 ± 0.06	−0.80 ± 0.26***	−0.99 ± 0.25***	−0.04 ± 0.24	0.02 ± 0.01*
Diet beverages	0.46 ± 0.52	0.33 ± 0.32	−0.05 ± 0.04	0.18 ± 0.48	0.11 ± 0.45	0.09 ± 0.15	0.05 ± 0.47	−0.04 ± 0.09	−0.02 ± 0.34	−0.15 ± 0.31	−0.62 ± 0.28**	−0.05 ± 0.02***
Bottled and flavored water	0.66 ± 0.44	0.11 ± 0.21	−0.05 ± 0.04	0.60 ± 0.39	0.28 ± 0.35	0.05 ± 0.14	0.15 ± 0.34	0.09 ± 0.07	0.01 ± 0.26	0.15 ± 0.24	−0.12 ± 0.23	−0.04 ± 0.01***
Dairy products (excluding milk)	−0.37 ± 0.23	0.03 ± 0.10	−0.04 ± 0.02*	−0.37 ± 0.21*	0.20 ± 0.20	0.14 ± 0.07**	−0.39 ± 0.20*	−0.03 ± 0.05	−0.03 ± 0.15	−0.14 ± 0.17	0.01 ± 0.14	0.00 ± 0.01
Meat, poultry, fish, and mixed meat dishes	−0.52 ± 0.28*	−0.08 ± 0.13	−0.05 ± 0.02**	−0.40 ± 0.25	−0.28 ± -1.07	0.11 ± 0.09	−0.41 ± 0.25	0.01 ± 0.04	−0.05 ± 0.18	−0.03 ± 0.18	0.19 ± 0.14	0.00 ± 0.01
Other protein sources	−0.06 ± 0.21	−0.06 ± 0.11	−0.03 ± 0.02	0.03 ± 0.20	−0.06 ± 0.22	0.15 ± 0.08*	−0.25 ± 0.22	0.05 ± 0.04	−0.04 ± 0.15	−0.01 ± 0.16	0.06 ± 0.15	0.00 ± 0.01
Grain products (excluding RTE desserts)	−0.19 ± 0.17	0.07 ± 0.10	−0.03 ± 0.02*	−0.23 ± 0.16	−0.19 ± 0.17***	0.09 ± 0.06	−0.25 ± 0.16	−0.03 ± 0.03	−0.01 ± 0.12	−0.15 ± 0.11	0.06 ± 0.10	0.00 ± 0.01
Fruits and vegetables	−0.38 ± 0.23	−0.08 ± 0.12	−0.05 ± 0.03**	−0.25 ± 0.22	−0.11 ± 0.24	0.10 ± 0.08	−0.16 ± 0.24	−0.05 ± 0.04	−0.26 ± 0.16	−0.29 ± 0.17*	−0.02 ± 0.15	−0.02 ± 0.01**
Fats, oils, sauces, and condiments	−0.42 ± 0.28	−0.05 ± 0.14	−0.01 ± 0.02	−0.36 ± 0.24	−0.22 ± 0.25	0.08 ± 0.08	−0.26 ± 0.23	−0.05 ± 0.04	0.15 ± 0.26	−0.27 ± 0.17	−0.01 ± 0.13	0.00 ± 0.01
Sweets and snacks	−0.07 ± 0.19	0.03 ± 0.10	−0.01 ± 0.02	−0.08 ± 0.16	−0.02 ± 0.17	0.12 ± 0.06*	−0.11 ± 0.15	−0.03 ± 0.04	−0.10 ± 0.12	−0.16 ± 0.12	0.01 ± 0.11	−0.01 ± 0.01
“Other” foods	−0.21 ± 0.27	0.12 ± 0.11	−0.03 ± 0.03	−0.30 ± 0.25	−0.20 ± 0.21	0.13 ± 0.09	−0.36 ± 0.21*	0.04 ± 0.05	0.20 ± 0.16	−0.48 ± 0.18***	−0.48 ± 0.16***	0.01 ± 0.01
Mixed dishes and soups	−0.19 ± 0.28	−0.07 ± 0.12	−0.04 ± 0.02*	−0.08 ± 0.25	−0.32 ± 0.21	0.05 ± 0.09	−0.40 ± 0.21*	0.02 ± 0.04	0.26 ± 0.16	0.10 ± 0.16	−0.19 ± 0.14	0.00 ± 0.01
All other beverages	−0.19 ± 0.17	0.07 ± 0.10	−0.03 ± 0.02*	−0.23 ± 0.16	−0.19 ± 0.17	0.06 ± 0.07	−0.05 ± 0.22	−0.02 ± 0.04	−0.28 ± 0.15*	−0.15 ± 0.11	0.06 ± 0.10	0.00 ± 0.01
Total beverages	−0.33 ± 0.22	−0.05 ± 0.10	−0.02 ± 0.02	−0.27 ± 0.20	0.04 ± 0.20	0.05 ± 0.07	0.03 ± 0.19	−0.04 ± 0.04	−0.28 ± 0.13**	−0.15 ± 0.13	−0.07 ± 0.12	−0.01 ± 0.01
Total foods	−0.20 ± 0.14	0.00 ± 0.07	−0.03 ± 0.02**	−0.17 ± 0.13	−0.13 ± 0.13	0.14 ± 0.07**	−0.39 ± 0.20*	−0.03 ± 0.05	−0.03 ± 0.15	−0.15 ± 0.09*	−0.05 ± 0.08	0.00 ± 0.00

¹Values are mean elasticities (defined as the percent change in amount purchased divided by the percent change in price) ± SEs; SEs were computed with use of bootstrapping for all foods/beverages purchased by fewer than 80% of households sampled. Estimates shown were computed with use of a 2-part marginal effects model (probit regression; OLS regression) for all foods/beverages purchased by fewer than 80% of households sampled. For all other foods/beverages, OLS regression alone was used. ***Significantly different from the null value, P < 0.01; **Significantly different from the null value, P < 0.05; *Significantly different from the null value, P < 0.10. OLS, ordinary least squares; RTE, ready-to-eat; SSB, sugar-sweetened beverage. University of North Carolina calculation based in part on data reported by Nielsen through its Homescan Services for all food categories, including beverages and alcohol for the 2009–2012 periods, for the US market. Copyright © 2015 The Nielsen Company.

Cross-price elasticities, defined here as the change in per capita purchases in grams divided by the change in price for another food/beverage, are also presented in Table 2. A complementary relation, denoted by a negative cross-price elasticity, indicates that increasing the price of one beverage decreases purchases of another food/beverage. Total SSBs were a complement to soft drinks (−0.75), juice drinks (−1.01), low-fat, low-sugar milk (−1.50), and meat, poultry, fish, and mixed meat dishes (−0.52). Juice drinks were a complement to low-fat, low-sugar milk (−1.28), whereas low-fat, low-sugar milk was a complement to 100% juice (−0.80). 100% Juice was also a complement to low-fat, low-sugar milk (−0.79). A positive cross-price elasticity indicates that increasing the price of one beverage increases purchases of another food/beverage. This is known as a substitution relation. Total SSBs were a substitute for sport and energy drinks (0.56), whereas >1% fat, low-sugar milk was a substitute for 100% juice (0.67).

Simulated taxes on SSBs (regular soft drinks, fruit drinks, and sport and energy drinks).

Table 3 shows adjusted purchases by weight (g/d) per capita for selected beverages, total beverages, and total foods (with no taxes), and the estimated change in purchases associated with increases in the prices of SSBs (regular soft drinks, juice drinks, and sport and energy drinks), and tax on SSBs and >1% fat and/or high-sugar milk. Increasing the price of SSBs by 10%, 15%, and 20% was associated with fewer purchases of juice drinks (range: −1.2 to −2.3 g/d per capita), and greater purchases of sport and energy drinks (range: 0.6–1.3 g/d per capita). There were no significant associations between increases in the price of SSBs with total weight (grams) purchases of beverages, foods, or food/beverages, although total purchases were predicted to decrease (range: −5.2 to −2.7 g/d per capita; P > 0.10).

TABLE 3.

Regression-adjusted mean purchases of >1% fat, low-sugar milk, total SSBs, non-SSBs, total beverages, and total foods, and predicted change in grams/kilocalories purchased per capita for select beverages associated with taxes of 10%, 15%, and 20% on SSBs, and on SSBs and >1% fat and/or high-sugar milk (n = 56,963)¹

		10% Increase in price		15% Increase in price		20% Increase in price
Change in quantity purchased	Regression-adjusted mean purchases (no taxes),2 g	SSBs, g	SSBs and >1% fat and/or high-sugar milk, g	SSBs, g	SSBs and >1% fat and/or high-sugar milk, g	SSBs, g	SSBs and >1% fat and/or high-sugar milk, g
Total SSBs	106 ± 1.8	−1.8 ± 2.5	−1.5 ± 3.1	−2.6 ± 3.6	−2.1 ± 4.6	−3.4 ± 4.7	−2.8 ± 6.0
Soft drinks	28.8 ± 2.0	−1.0 ± 1.2	0.5 ± 1.5	−1.4 ± 1.8	0.8 ± 2.3	−1.9 ± 2.4	1.0 ± 3.0
Sport and energy drinks	7.5 ± 3.2	0.6 ± 0.3**	−0.1 ± 0.3	0.9 ± 0.4**	−0.2 ± 0.4	1.3 ± 0.5**	−0.2 ± 0.5
Juice drinks	24.0 ± 1.6	−1.2 ± 0.8*	−1.0 ± 0.8	−1.8 ± 1.1*	−1.5 ± 1.2	−2.3 ± 1.5*	−1.9 ± 1.5
Total >1% fat and/or high-sugar milk	89 ± 1.6	0.5 ± 1.0	−6.2 ± 2.4*	0.8 ± 1.5	−8.9 ± 3.5*	1.0 ± 1.9	−11.4 ± 4.6*
>1% fat, high-sugar milk	3.5 ± 1.9	0.0 ± 0.1	−0.1 ± 0.1*	0.0 ± 0.1	−0.2 ± 0.1*	0.0 ± 0.2	−0.3 ± 0.2*
>1% fat, low-sugar milk	86.1 ± 1.6	0.0 ± 2.7	−6.0 ± 8.2	0.0 ± 4.0	−8.6 ± 12.4	0.1 ± 5.2	−11.1 ± 16.6
Low-fat, high-sugar milk	1.8 ± 1.6	0.0 ± 0.0	0.0 ± 0.0	0.0 ± 0.0	0.0 ± 0.0	0.0 ± 0.0	−0.1 ± 0.1
Low-fat, low-sugar milk	7.4 ± 1.9	−0.1 ± 0.4	1.0 ± 0.5*	−0.1 ± 0.5	1.6 ± 0.8*	−0.1 ± 0.7	2.2 ± 1.1*
100% Juice	8.3 ± 1.8	0.1 ± 0.3	0.5 ± 0.6	0.2 ± 0.4	0.8 ± 0.8	0.3 ± 0.6	1.1 ± 1.1
Diet beverages	12.3 ± 2.2	0.6 ± 0.6	−0.4 ± 0.8	0.8 ± 0.9	−0.5 ± 1.2	1.1 ± 1.2	−0.7 ± 1.6
Bottled and flavored water	28.3 ± 13.6	0.6 ± 1.1	−0.1 ± 1.0	0.8 ± 1.6	−0.2 ± 1.5	1.1 ± 2.2	−0.3 ± 1.9
Other (non-SSB) beverages	340 ± 1.4	2.6 ± 6.6	−2.5 ± 12.8	3.8 ± 9.7	−3.6 ± 18.9	5.0 ± 12.7	−4.7 ± 24.9
Total beverages	434 ± 1.3	−3.2 ± 7.4	−2.7 ± 10.2	−4.6 ± 10.9	−3.9 ± 15.1	−6.0 ± 14.3	−5.1 ± 19.8
Dairy products (excluding milk)	30.8 ± 1.4	−0.2 ± 0.0	−0.7 ± 48.4	−0.2 ± 0.0	−1.0 ± 4.7	−0.3 ± 0.0	−1.2 ± 0.5
Meat, poultry, fish, and mixed meat dishes	20.7 ± 1.5	−0.5 ± 0.0	−0.4 ± 93.1	−0.7 ± 0.0	−0.6 ± 92.9	−0.9 ± 0.0	−0.7 ± 6.2
Other proteins	9.4 ± 1.4	0.2 ± 0.2	−0.2 ± 0.2	0.2 ± 0.3	−0.4 ± 0.3	0.3 ± 0.4	−0.5 ± 0.5
Grain products (excluding RTE desserts)	64.0 ± 1.2	−0.2 ± 0.0	−0.9 ± 105.7	−0.3 ± 0.0	−1.2 ± 105.8	−0.4 ± 0.0	−1.6 ± 13.9
Fruits and vegetables	63.2 ± 1.4	−0.5 ± 1.2	−1.0 ± 1.4	−0.8 ± 1.8	−1.5 ± 2.1	−1.0 ± 2.4	−1.9 ± 2.7
Fats, oils, sauces, and condiments	28.4 ± 4.7	−0.2 ± 0.6	−0.4 ± 0.5	−0.3 ± 0.9	−0.5 ± 0.8	−0.4 ± 1.2	−0.7 ± 1.0
Sweets and snacks	66.9 ± 1.3	0.4 ± 1.0	0.4 ± 1.3	0.7 ± 1.4	0.5 ± 1.9	0.9 ± 1.8	0.7 ± 2.4
“Other” foods	6.7 ± 2.0	0.1 ± 0.2	0.0 ± 0.2	0.1 ± 0.2	−0.1 ± 0.3	0.1 ± 0.3	−0.1 ± 0.4
Mixed dishes and soups	40.6 ± 1.5	0.2 ± 0.9	−1.2 ± 0.8	0.3 ± 1.3	−1.8 ± 1.2	0.4 ± 1.7	−2.3 ± 1.6
Total foods	331 ± 1.2	−0.1 ± 3.8	−3.0 ± 9.1	−0.1 ± 5.5	−4.3 ± 13.4	−0.1 ± 7.2	−5.7 ± 17.6
Total foods/beverages	766 ± 1.2	−2.7 ± 8.7	−7.4 ± 21.4	−4.0 ± 12.8	−10.8 ± 31.6	−5.2 ± 16.8	−14.0 ± 41.5

¹Values are means ± SEs. **Significantly different from mean value with no tax, P < 0.05; *Significantly different from mean value with no tax, P < 0.10. FPL, federal poverty level; RTE, ready-to-eat; SSB, sugar-sweetened beverage.

²Models were adjusted for household composition (number of household members by gender and age: 2–5 y; 6–11 y; 12–18 y; ≥19 y); household income as a percent of the FPL (≤100% FPL; >100–130% FPL; >130–185% FPL; >185–400% FPL; and >400% FPL), education level (highest level of education completed by head of household), race/ethnicity, and quarterly unemployment rate by market, year, and quarter.

Simulated taxes on SSBs and >1% fat and/or high-sugar milk.

Table 3 also shows the predicted changes in purchase (g/d) per capita associated with simultaneous price increases of 10%, 15%, and 20% on SSBs and >1% fat and/or high-sugar milk. Increases in the prices of SSBs and >1% fat and/or high-sugar milk were associated with fewer purchases of >1% fat, low-sugar milk (range: −10.2 to −5.5 g/d per capita), meat, poultry, fish, and mixed meat dishes (range: −2.1 to −1.1 g/d per capita), and increased purchases of sport and energy drinks (range: 0.8–1.6 g/d per capita). Neither 10%, 15%, or 20% increases in the prices of SSBs and >1% fat and/or high-sugar milk was significantly related to weight or caloric purchases of individual foods or beverages, total foods, total beverages, or total foods/beverages (P > 0.10). However, total purchases of foods/beverages were predicted to decrease (range: −20.7 to −10.9 g/d per capita; P > 0.10).

Simulated beverage taxes and total annual caloric purchases.

Figure 2 shows the estimated annual associations between total kilocalories purchased from foods and beverages and increases in the prices of SSBs alone, or increases in the prices of both SSBs and >1% fat and/or high-sugar milk. Price increases of 10% to 20% on SSBs alone were associated with nonsignificant decreases (P = 0.79) in annual total kilocalories purchased per capita of between 1177 and 2228 kilocalories. Ten to 20% increases in the prices of both SSBs and >1% fat and/or high-sugar milk were associated with nonsignificant decreases in annual total kilocalories purchased of between 3287 and 6245 kilocalories (P = 0.35).

FIGURE 2.

Change in mean grams purchased annually per capita with taxes of 10%, 15%, and 20% on SSBs (regular soft drinks, juice drinks, and sport and energy drinks), or on SSBs and >1% fat and/or high-sugar milk. Values are given as mean annual change in kilocalories purchased per capita among US households with a preschool child who participated in the Nielsen Homescan Panel, 2009–2012. University of North Carolina calculation based in part on data reported by Nielsen through its Homescan Services for all food categories, including beverages and alcohol for the 2009–2012 periods, for the US market. SSB, sugar-sweetened beverage. Copyright © 2015 The Nielsen Company.

Discussion

In this paper, we used simulated price increases as a proxy for “taxes” to examine the association between “taxes” of 10%, 15%, and 20% on SSBs and food and beverage purchases among households with a child 2–5 y of age. We compared this model with one in which price increases were simulated for both SSBs and >1% fat and/or high-sugar milk, and contrasted associations with kilocalories purchased from SSBs, non-SSBs, total beverages, and total foods between “tax” models. Increases in the prices of SSBs alone were significantly related with fewer purchases of juice drinks. In contrast, concomitant increases in the prices of SSBs and >1% fat and/or high-sugar milk were associated with fewer purchases of >1% fat low-sugar milk, but were not associated with significant reductions in purchases of any SSB. In all models, there were no significant associations between price increases (on SSBs or SSBs and >1% fat and/or high-sugar milk) and kilocalories or grams purchased from total beverages, total foods, or total foods and beverages. Although not statistically significant, total kilocalories purchased was expected to decrease in both models but to a greater extent when the prices of both SSBs and >1% fat and/or high-sugar milk were increased.

Regardless of its fat or sugar content, milk has some redeeming nutritional qualities, like calcium and vitamin D (38). Moreover, despite prevailing recommendations to limit intakes of >1% fat and/or high-sugar milk in children ≥2 y of age (37, 49–54), the relation between intakes of high-fat milk and overweight in preschool children is unclear (55). In contrast, there is a general consensus that SSBs have little nutritional value (38) and that their consumption may promote excess weight gain in children (56). Thus, although there appear to be benefits from both tax models, the relation between increases in the prices of SSBs alone and beverage purchases were marginally more favorable than those associated with a combined tax.

Although this is the first study to focus exclusively on preschooler households, like others we found that increasing the prices of SSBs would shift purchases away from juice drinks and toward purchases of 100% juice. Finkelstein et al. (25), who also used data from the Nielsen Homescan Panel (2006), reported that a 20% tax on SSBs (regular soda, fruit drinks, and sport drinks) was associated with a reduction in purchases of juice drinks and soft drinks and substantial increases in purchases of fruit juices (25). This finding was also supported by Smith et al. (28) who reported that a 20% tax on SSBs (regular soft drinks, juice drinks, sport and energy drinks, and powdered mixes with added sugars) would simultaneously decrease purchases of SSBs while increasing purchases of juices (28). We did not, however, observe a significant reduction in purchases of soft drinks, although point estimates were in the expected direction.

To our knowledge, ours is the first study to simulate simultaneous increases in the prices of both SSBs and >1% fat, high-sugar milk. Thus, there are no studies with which to compare our results. However, our observed own-price elasticities for the additional beverages included in the combined tax model are consistent with previous reports. For example, we observed an own-price elasticity for juice drinks of −1.01, whereas several studies also using Homescan data have reported values in the range of −1.19 to −1.02 for juices and juice drinks (27, 28, 30). Similarly, we observed an own-price elasticity for low-fat, low-sugar milk of −0.79, whereas others have reported elasticities for 1% and skim milk in the range of −0.90 to −0.40 (57–59). We found an own-price elasticity for >1% fat, low-sugar milk of −0.65, whereas others have reported values ranging from −0.90 to −0.43 for 2% and whole milk (57–61). Nonetheless, these prior studies do not provide a one-to-one comparison because they did not discern between high- and low-sugar milk. Notably, we observed a smaller own-price elasticity for soft drinks than has been previously reported (62). However, studies using demand systems have predominated the literature of late (26–28, 30, 48), and elasticity estimates from these studies have tended to be larger in magnitude than those observed when 2-part models are used (25, 43). Finkelstein et al. (25), who also used both a 2-part model and Homescan data, reported own-price elasticities for soft drinks ranging from −0.476 ± 0.281 to −0.908 ± 0.072. Duffey et al. (43), who also used a 2-part model but used data from the Coronary Artery Risk Development in Young Adults Study, reported an own-price elasticity estimate for soft drinks of −0.712 ± 0.183. By comparison, our own-price elasticity estimate for soft drinks of −0.15 ± 0.21 was not significantly different from zero, but within margin of error of these prior estimates. We suspect that insufficient power, as a result of significant measurement error and sample restriction (households with a single preschool child), likely also played a role. Moreover, preschool children consume fewer kilocalories from soft drinks than older children and adults (97 kcal/d vs. 301 kcal/d for children 12–19 y of age) (63, 64), who have been the focus of most prior studies.

We also examined the potential for targeted beverage “taxes” to increase intakes of other foods high in fats and/or sugar (26). We found no evidence, however, that either beverage “tax” scenario (SSBs alone, or SSBs and >1% fat and/or high-sugar milk) would significantly influence total kilocalories purchased. This was true for simulated price increases of 10%, 15%, and 20%. Notably, 2 prior studies, each comprising a general sample of US households, reported that a 20% tax on SSBs was predicted to decrease total kilocalories purchased by −17.9 kcal/d per capita (26) and by −24.3 kcal/d per capita (25), respectively. In comparison, we found that a 20% tax on SSBs was associated with purchasing −22 kcal/d per capita, although this result did not reach statistical significance. Notably, our sample was limited to US households with a single preschool child, and we performed statistical adjustments to best scale our estimates relative to a preschool child. Thus, differences in our sample offer one potential explanation for the discrepant finding because our sample was limited to households with a single preschool child, whereas most prior studies examine these relations in a general sample of households.

Finally, as the predominance of studies examine taxes of 20% or more (25–28, 30), we sought to determine whether price increases (on SSBs, or SSBs and >1% fat and/or high-sugar milk) of <20% (10% and 15%) were significantly associated with purchases of SSBs and/or >1% fat, high-sugar milk. It has been previously suggested that taxes <20% would not have an appreciable influence on consumer behavior (65–67). However, the few prior studies that examined how beverage taxes of <20% influence consumer behavior (66, 67) used state-level soft drink sales taxes—which tend to be small in magnitude—to explore this relation. Moreover, because sales taxes are not typically reflected in shelf price, they are unlikely to influence consumer behavior (68). In contrast, our “tax” models assume an excise tax for which 100% of the tax is transferred to the shelf price, which is in keeping with previous works (26, 27, 29, 30). We found that price increases as little as 10% on SSBs were significantly associated with fewer purchases of juice drinks and greater purchases of 100% juice and low-fat, low-sugar milk. Increases in the prices of SSBs were also nonsignificantly related to fewer total purchases of foods and beverages by weight. However, such changes, even with a 20% increase in price, were small in magnitude (<15 g/d per capita). Although it is possible that the actual effects of a tax may be larger than those we observed, because Homescan does not capture all food/beverage purchases (35), our findings suggest that taxes of 20% or more would be needed for more meaningful changes in food/beverage purchases among US households with a preschool child.

There are several key limitations to our study. Foremost, our findings reflect associations, rather than causal relations, because the outcomes (amount purchased) and primary exposures (prices paid) were ascertained at the same point in time. Additionally, we are unable to directly determine which foods or beverages are consumed by who in each household because purchases are measured at the household—rather than individual—level. However, we have undertaken several steps to best estimate per capita purchases. We included only households with a single preschool child, in an effort to minimize heterogeneity in household composition. We also controlled for household composition in all of our statistical models, including number of household members by gender and age (0–2; 2–5; 6–11; 12–18; ≥19 y). Nonetheless, inferences from our findings are limited to households with a preschool child.

A further limitation of the Homescan data is that foods and beverages without barcodes—including fresh produce and meats, as well as foods purchased at restaurants, school cafeterias, or child care centers—tend to be poorly reported (or not reported altogether) (35). Thus, these items were excluded from the analysis. Notably, these items are nontrivial, with fruits, vegetables, and meat expenditures combined comprising roughly 17% of total household food expenditures among US Homescan respondents (69). Additionally, US preschool children obtain ∼27% of total calories from sources outside the home (64). Nonetheless, the principal aim of this paper was to estimate the association between taxes on high-sugar and/or >1% fat beverages and beverage purchases. Because beverages, along with consumer-packaged foods, many of which are key sources of fats and sugars (70), are well-represented in Homescan (71), we are confident that these data allow us to examine our research aim.

Our study provides further evidence that a tax on beverages high in sugar and/or fat may be associated with favorable shifts in food/beverage purchases among US households with a preschool child. We also found no evidence that either a tax on SSBs alone or a tax on both SSBs and >1% fat and/or high-sugar milk would increase purchases of other foods/beverages or total kilocalories. Moreover, targeted beverage taxes as little as 10% could shift purchases away from beverages high in fats and/or added sugars. However, observed changes even at 20% tax rates were small, suggesting that taxes of 20% or more on SSBs may be needed to appreciably change food/beverage purchases among US households with a preschool child.