A sales tax is better at promoting healthy diets than the fat tax and the thin subsidy. Zarko Kalamov. Health Economics, 2020;1–14, DOI: 10.1002/hec.3987
Abstract: We analyze how a sales tax levied on all food products impacts the consumption of healthy food, unhealthy food, and obesity. The sales tax can stimulate the consumption of healthy meals by lowering the time costs of food preparation. Moreover, the sales tax lowers obesity under more general conditions than a tax on unhealthy food (fat tax) and a subsidy on healthy food (thin subsidy). We calibrate the model using recent consumption and time use data from the US. The thin subsidy is counterproductive and increases weight. While both the sales tax and the fat tax mitigate obesity, the former imposes a lower excess burden on consumers.
Keywords: fat tax, obesity, sales tax, thin subsidy
JEL: D11; I12; I18; H31; H51
1 INTRODUCTION
Many countries tax unhealthy foods to address the obesity epidemic: Chile, France, Ireland, Mexico, and the UK among others tax sugar-sweetened beverages (SSBs), Finland taxes sweets and non-alcoholic beverages, while Hungary taxes food products with health risks. However, these policies may be ineffective or even counterproductive in reducing obesity, as consumers may substitute to untaxed unhealthy food alternatives (Schroeter, Lusk, & Tyner, 2008). To minimize this problem, governments can broaden the tax base (Finkelstein, Zhen, Nonnemaker, & Todd, 2010; Harding & Lovenheim, 2017; Miao, Beghin, & Jensen, 2013). Therefore, nutrient taxes (such as a sugar tax or a fat tax) are more effective than product taxes (such as a tax on SSBs). This article is the first to propose a sales tax on all food products as an instrument that promotes healthy diets. A sales tax may stimulate healthy consumption by lowering the opportunity cost of cooking time. Moreover, a sales tax imposes a low excess burden on consumers, measured both per kcal reduction in consumption and per dollar of tax revenues. We model a representative consumer in a model akin to that of Yaniv, Rosin, and Tobol (2009) and take explicitly into account the higher opportunity cost in terms of time of healthy consumption. The individual chooses between consumption of healthy and unhealthy food and has a fixed out-of-work time constraint, which she can spend on cooking meals and leisure. The individual consumes unhealthy meals away-from-home (defined as food prepared away-from-home), and their preparation is not time-consuming. Healthy food is produced at home using both time and ingredients.
We show that the sales tax lowers the opportunity cost of time in food preparation and may thus stimulate healthy consumption. If the elasticity of substitution between healthy and unhealthy food is sufficiently high, the sales tax exerts a positive effect on the demand for healthy meals. A tax on unhealthy foods (called for simplicity a fat tax) and a subsidy to healthy consumption (called a thin subsidy) have qualitatively similar effects on the demand for healthy and unhealthy food. Moreover, we show that the policy, which reduces obesity under the least restrictive conditions, is a positive sales tax in the absence of a fat tax and a thin subsidy.
Our model builds on the empirical observation that home-cooked meals are healthier than away-from-home food. Compared to the consumption of away-from-home meals, intake of home-cooked food is associated with higher intake of fiber, iron, and calcium; lower intake of fat, sodium, and cholesterol; lower calorie density (Guthrie, Lin, & Frazao, 2002; Lin & Frazao, 1997; Lin & Frazao, 1999) and lower weight (Chou, Grossman, & Saffer, 2004; French, Harnack, & Jeffery, 2000). Furthermore, a higher frequency of food preparation raises the consumption of fruits and vegetables and lowers the intake of fat, SSBs and fast-food (Larson, Perry, Story, & Neumark-Sztainer, 2006; Laska, Larson, Neumark-Sztainer, & Story, 2012; McLaughlin, Tarasuk, & Kreiger, 2003; Monsivais, Aggarwal, & Drewnowski, 2014; Wolfson & Bleich, 2015b). The positive dietary impact of frequent cooking occurs irrespectively of the weight-loss intentions of individuals (Wolfson & Bleich, 2015a). Kolodinsky and Goldstein (2011) estimate that ten additional minutes of cooking time lower BMI by 0.13 points. Zick, Stevens, and Bryant (2011) find that ten minutes food preparation lower the BMI of women by 0.17 points and do not affect the BMI of men, who, however, are a small share of the meal preparers in their data.
In Section 4, we calibrate the model according to recent U.S. consumption and time allocation data. We estimate the home production function, the dietary characteristics of at-home- and away-from-home-food and the time allocation decisions such that they match data from the National Health and Nutrition Examination Survey (NHANES) 2009-2010 (CDC, 2010) and the American Time Use Survey (ATUS) 2010 (BLS, 2010). The simulated model makes predictions regarding the own- and cross-price elasticities of healthy and unhealthy food that match existing empirical data on these elasticities.
We find in the benchmark case of our simulation that a 10% sales tax lowers away-from-home consumption of men and women by 7.9% and 4.8%, respectively. On the other hand, men's at-home consumption declines by just 0.01% and that of women rises by 0.01%. Hence, the intake of at-home food remains almost unaffected. As a result, men lose 4% of their weight and women 1.8%. On the other hand, a 10% fat tax induces a stronger substitution of at-home for away-from-home consumption. Consequently, it leads to a lower weight decrease. Moreover, the excess burden of the sales tax, measured using the compensating variation, is lower than that of the fat tax, irrespective of whether we calculate the excess burden as the welfare loss per dollar of tax revenues or per reduction in calorie intake. Furthermore, the sales tax's excess burden is also small in absolute value. It equals around 1 − 1.7 cents per dollar of tax revenues and 2 − 2.9 cents per 100 kcal reduction in consumption. Lastly, an introduction of a thin subsidy stimulates strong substitution of healthy for unhealthy food, such that weight slightly increases following the subsidy's implementation. The article from Yaniv et al. (2009) is most related to our paper. They analyze the fat tax and thin subsidy under consideration of the time costs of home-food preparation. Neither policy instrument is unambiguously obesity-reducing because of indeterminate substitution effects on away-from-home food consumption and physical activity.
Bishai (2015) considers the welfare implications of a tax plus subsidy system that raises (lowers) the price of unhealthy (healthy) nutrients but leaves the final product prices unchanged. He shows that such a system may improve welfare.
There is strong empirical evidence that a fat tax does not necessarily reduce weight, because of a possible switch to untaxed alternatives (Chouinard, Davis, LaFrance, & Perloff, 2007; Fletcher, Frisvold, & Tefft, 2010b; Schroeter et al., 2008; Zhen, Finkelstein, Nonnemaker, Karns, & Todd, 2014). Schroeter et al. (2008) estimate that a tax on food-away-from-home and a subsidy on fruit and vegetables may increase obesity. Zhen et al. (2014) show that a tax on SSBs can reduce their consumption at the cost of higher intake of fat and sodium. Furthermore, Fletcher, Frisvold, and Tefft (2010a), Fletcher et al. (2010b) find empirical support that such taxes have a significant effect on SSB consumption, but an insignificant effect on weight. Jeffery, French, Raether, and Baxter (1994) and French et al. (1997) find a significant short-term impact of subsidizing fruits and salads at university and high school cafeteria, which vanishes after the removal of the subsidy. [...]
5 CONCLUSIONS
This paper has compared three different policy instruments, which can be used to address the problem of rising obesity levels: a fat tax (levied on food-away-from-home), a thin subsidy (levied on groceries that enter home food preparation) and a sales tax on all food items. First, we show that a sales tax may stimulate time-intensive healthy consumption by lowering the opportunity cost of time spent on food preparation. Therefore, it may exert a positive effect on the demand for healthy meals. If healthy and unhealthy meals are perfect substitutes, then all three policy instruments have the same qualitative impact on the consumer's demand: they reduce the consumption of unhealthy meals and raise the consumption of healthy food. Second, the policy which reduces obesity under the most general conditions is a sales tax in the absence of the fat tax and the thin subsidy.
A calibration of the model shows that the sales tax mitigates obesity at the lowest welfare cost for consumers. Furthermore, the deadweight cost of the sales tax is small is absolute value. It imposes an excess burden of less than 2 cents per dollar of tax revenues and 3 cents per 100 kcal reduction in the calorie-intake.
Our results open ample opportunities for future research. While this article extends the model of Yaniv et al. (2009) to include a general elasticity of substitution between food at-home and away-from-home and the possibility of non-food purchases, we do not consider the choice of physical exercise. Including it may produce further interesting results on the effects of a sales tax.
Additionally, this article hightlights the role of the price of time in consumption choices. It explains the higher time spent cooking by women through lower opportunity cost of leisure. Gender differences in the price of time are likely caused by the gender wage gap, as documented by Zick et al. (2011). These authors use wage regressions from the March Supplement of the Current Population Survey (CPS) and find the opportunity cost of time for men and women to be 20.57 $∕hour and 16.84 $∕hour, respectively. The gender wage gap is also likely to contribute to unequal distribution of cooking time in non single-adult households. Moreover, closing of the gap is likely to affect this distribution. Hence, more research is necessary to analyze how different opportunity costs of time and the elimination of these differences may affect household production.
Moreover, future work should compare the sales tax to nutrient-specific taxes such as a tax on sugar or fat content. The sales tax may be more efficient in promoting healthy diets as it targets several unhealthy nutrients at once by lowering away-from-home consumption. Furthermore, a comparison to the tax plus subsidy system of Bishai (2015) along the same lines is necessary.
This paper has contributed to the literature by emphasizing the time costs of healthy consumption and showing their importance for the optimal policy design. Future research should focus on analyzing other policies that lower the opportunity cost of home food preparation or provide other incentives for cooking. Two public health programs that have already been implemented in the US are the Supplemental Nutrition Assistance Program (SNAP) Healthy Incentives Pilot which provides financial incentives for the purchase of fruits and vegetables and the Women, Infants, and Children (WIC) Farmer's Market Nutrition program which issues coupons to participants to buy foods from farmers (Smith, Ng, & Popkin, 2013). An evaluation of the effects of these programs on the participants' cooking habits is an important research agenda.
Taxing all food products may be regressive, as poor households spend a larger proportion of their income on food relative to rich households. As a result, an important agenda for future research is to quantify this effect and analyze how governments should spend the tax receipts, such that the tax becomes less regressive. Additionally, it is well-known that a fat tax is also regressive (see, e.g., Chouinard et al. (2007)). Therefore, future work should compare the regressivity of a fat tax to that of a sales tax.
Our research highlights the importance of including social conformity in the study of political phenomena. Recent research in political psychology has focused on the Big Five personality traits (see especially Mondak et al. 2010) but our study suggests that we should go beyond these personality traits. Politics is very much about collective decision-making, so there is an underlying tension between the desire for personal autonomy and the need for social norms that are respected and followed by everyone in the community. Citizens strike a different balance between these two considerations, and this is bound to shape their behavior.
We have combined a survey experiment conducted in Britain and a lab experiment performed in Canada to test our hypothesis. Both studies produced remarkably similar findings: in both instances, people who score higher on the social conformity scale are more prone to vote (abstain) when they know that most other people vote (abstain) while those who score low on the conformity scale have exactly the opposite reaction. If a predicted high turnout makes conformists more inclined to vote and if there are many conformists in the electorate, then a relatively high turnout would be an equilibrium.
The implications of our study hinge in good part of the distribution of individuals across the conformity scale in a given society at a given point in time. To the best of our knowledge there are no time-series or cross-section data about the distribution of conformity across societies or over time. The data that we have collected in Britain (see Fig. A1 in Online Appendix A1) suggest a relatively normal distribution, with the mean (7.52 on a 0 to 17 scale) indicating a slight majority of non-conformists. The point remains that, at least in our representative poll of British society, there are many people with conformist leanings.
We therefore end with a call for more attention to be paid to an individual’s social conformity. A huge literature exists in social psychology about the role of conformity, yet little research has been devoted to its impact in political life. Our study suggests that it is a crucial variable in the decision to vote or abstain. There are good reasons to believe that it shapes other political phenomena such as the decision to participate in demonstrations or to engage in strategic or bandwagon voting. It makes sense to assume that many people pay attention to information about what others in the community are likely to do when deciding whether to vote or abstain. Political scientists need to integrate such considerations into their models and analyses.
Our study also raises important questions about the relationship between social conformity and other factors that may affect voter turnout. One such question pertains to the relationship between social conformity and sense of civic duty. Are conformists more prone to believe that they have a moral obligation to vote? Are both attitudes shaped by personality traits? Another set of questions is about the relationship between social conformity and social pressure. We would expect conformists to pay attention to social pressure. But what kind of social pressure? In this study, we have examined how conformists react to information about aggregate turnout. But what happens when conformists are exposed to conflicting information, if/when for instance they learn that turnout in the country is going to be high but that most friends/relatives are going to abstain? These are big questions about which we have no clear answer.