Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Hunger in the United States wikipedia , lookup
Food safety wikipedia , lookup
Food and drink prohibitions wikipedia , lookup
Human nutrition wikipedia , lookup
Obesity and the environment wikipedia , lookup
Food studies wikipedia , lookup
Food coloring wikipedia , lookup
Taxing Animal Products: Protein Demand under Environmental Pressure and Social Impact in France F. Caillavet, A. Fadhuile, V. Nichèle INRA-UR1303, ALISS Selected Paper prepared for presentation at the Agricultural Applied Economics Associations 2014 AAEA Annual Meeting, Minneapolis, MN, July 27-29, 2014. Copyright 2014 by F. Caillavet, A. Fadhuile, and V. Nichèle. All rights reserved. Readers may make verbatim copies of this document for non-commercial purposes by any means, provided that this copyright notice appears on all such copies. Taxing Animal Products: Protein Demand under Environmental Pressure and Social Impact in France F. Caillavet1 , A. Fadhuile1 , V. Nichèle1 May 2014 Preliminary version. Please do not cite or quote without permission of the authors. Abstract Europe committed to reduce greenhouse gas emissions (GHG) by 40 % by 2030 from 1990 levels. Food emits about 30% of all GHG, the major toll arising from animal products (half of food GHG). This urges the necessity of public actions to encourage sustainable diets. Food policy is now at the double stake of preserving environment and improving health. To implement a public policy combining environmental and nutritional issues in a sociallyconscious framework, a food demand study and the potential substitutions between foods is necessary. This article aims at offering a solid base for such policy decisions. Which are the food groups more suitable for a price change? Where are the more disparities in price responsiveness among income classes? Are own-price effects the only relevant? Do crossprice effects matter? To study food demand, we estimate an EASI demand system. It is based on a pseudo panel of 8112 observations constructed from Kantar panel data (1998-2010). It registers French households purchases for food-at-home. We add the nutritional content and Greenhouse gas emission related to foods through Life Cycle Analysis. For 21 food groups, built according to their environmental and nutritional characteristics, we run expenditure and price elasticities. Based on these results, two taxation scenarios are implemented. For each we increase by 20% the prices of food categories with most adverse effects on (1) environment only (ENV) and (2) both environment and health (ENV-NUT). The ENV scenario induces more reductions in environment impact than the ENV-NUT scenario, in particular for SO2 emissions. A greatest impact is observed for lower-average income households and with a household head less than 30 years old. However, undesirable nutritional effects lead to consider the necessity of a trade-off between environment and nutrition. Our conclusions find that this trade-off is not so costly (-18% in terms of CO2 ). Moreover, our results give new insights for targeting public policies toward the youngest households which we find more sensitive to prices and which are at the beginning of the consumption life-cycle. Keywords: EASI demand system, food purchases, socioeconomic inequalities, public policy. JEL Classification: C35, D12, Q15. 1 INRA-UR1303, ALISS Ivry-sur-Seine. This work was supported in part by FERRERO Cie and by the French National Research Agency under the OCAD project (Offer and consume a sustainable diet). Our acknowledgements go to O. Allais for his comments and N. Guinet for his data assistance. Corresponding author: [email protected] 1 1 Introduction Food consumption is estimated to be responsible for 30% of Greenhouse gas emission (GHG) in Europe. To fill the European commitment to reduce GHG emissions by 40% before 2030, changes of diet seem unavoidable. In a global context of increasing pathologies related to nutrition, food policy is now at the double stake of preserving environment (hereafter sustainability) and improving health (here addressed through nutritional objectives). Furthermore, nutrition and health show a strong social gradient (Mackenbach et al. (2008)). Socioeconomic disparities in the purchase basket lead also to differentiated environmental impacts (Boeglin, Bour, and David (2012)). Encouraging an environment-friendly diet through public action must take into account nutritional and social consequences, and food policies should be implemented with those three combined objectives. Can economic incentives drive environmental sustainability and healthier diets ? Literature and some real life experiences of price policies for health purposes have already addressed some key points: would taxing less healthy products or lowering market prices/subsidizing healthier products increase consumption of desired more healthful products ? Some examples of increasing VAT on unhealthy products (fat, junk food) have been implemented in some countries (France, Allais, Bertail, and Nichèle (2010); Denmark, Smed, Jensen, and Denver (2007); UK, Briggs et al. (2013)) with controversial results in terms of efficiency on the diet and population targeted. In the environmental field, the question whether similar tools could be implemented to encourage both environmental-friendly and healthy food choices raises at least two issues. First, the food groups targeted for environmental reasons may be different than those targeted for health reasons. Second, a price policy may have divergent rationale for implementation, due to price formation: in the environmental framework, price increases result from scarcity due to rarefac1 tion of resources more than from a regulatory policy decision. Besides, a price intervention could be more adequate at the level of producers in case of environment than in the health one (Capacci et al. (2012)). However, the perspective of inducing a more favourable diet for both health and environment through consumer prices is not irrelevant. Regarding the first issue in the literature, i.e. designing sustainable food policies facing the complexity environment/nutrition, most papers consider policy scenarios or simulate a change in diet dealing with a reduction of meat consumption, since the major toll arises from animal products (McMichael et al. (2007)). In particular, recent studies in UK evaluate the impact of alternative diets lower in red meat and processed meat (Aston, Smith, and Powles (2012)) or different diet scenarios where the largest potential reduction in GHG emissions is achieved by eliminating meat from the diet (35% reduction), followed by changing from carbon-intensive lamb and beef to less carbon-intensive pork and chicken (18% reduction) (Hoolohan et al. (2013)). Scarborough et al. (2012) study the health perspectives of three dietary scenarios based on GHG emissions and finds a potential for substantial improvements. However, less compatibility in objective is not always found. In the French case, Vieux et al. (2013) results are controversial. The impact of different meat reduction scenarios is modest and they emphasize adverse interactions between health and environment aims. Keeping calorie intake constant, when substituting fruit and vegetables for meat in a healthier perspective, GHG emissions may even increase (Masset et al. (2014)). A more global work at the European level considers the consequences of six alternative diets consisting of a 25% or 50% reduction in the consumption of beef, dairy, pork, poultry and eggs, compensated by a higher intake of cereals (Westhoek et al. (2014)). Note that the evaluation of environmental impact is realized mostly through a single indicator (CO2 ), which is quite restrictive. Similarly to health where composite indicators have been built such as the Healthy Eating Index (Drewnowski et al. (2009)), some literature environmental index combining various 2 indicators. These scenarios simulating changes in diets have methodological drawbacks: most of them do not consider a full consumption framework taking into account substitution behaviours. In the French case, both papers cited above modelled various a priori hypothesis of substitutions between food groups, which were not based on the estimation of elasticities from a demand system. One reason for this is that they do not deal with the issue of how to obtain this change of diet. Indeed, when we come to the second issue in the literature, i.e. implementing a price policy at the consumer level for lowering GHG emissions, works are rather scarce. The taxation of food products with higher GHG emissions through higher value-added taxes (VAT) on meat and dairy products, or CO2 emissions level taxes (Edjabou and Smed (2013)), evidences the ambiguity of increasing the price of healthy foods such as low fat sources of animal proteins, for example milk. While waiting for guidelines which would combine health and environmental objectives, the compatibility issue can only be driven by trade-off insights. A third issue deals with heterogeneity of consumption, meaning in particular different consumption patterns and price sensitivity according to socioeconomic characteristics. A recent paper on Danish data points out that income and education gradients in lifestyle choices vary with age (Ovrum, Gustavsen, and Rickertsen (2014)). Consumption of key products for health consequences such as fruits and vegetables show a widening income gradient with age till 70 years. In the French context, it has been proven to show strong age and generation effects (Hébel and Recours (2007)). Chancel (2014) points out the importance of the generational dimension in consumption patterns and consequent environmental footprint. Some works emphasize that strategies to change meat eating frequencies and meat portion sizes appeal to different segments of consumers which should be addressed in terms of their own preferences. A point of great 3 interest remains in testing the existence of a higher price sensitivity for low income households in order to evaluate the potential of food policies to reduce socioeconomic inequalities. When considering a tax on food prices which involves by nature a regressive content, a specific issue for low income households should take into account food security and the eventual need of a program assistance or compensatory mechanism. In the perspective of implementing an economic policy combining environmental and nutritional issues in a socially-conscious framework, a food demand study and the potential substitutions between foods is necessary. This article aims at offering a solid base for such policy decisions, which could aim at reducing the carbon content of food purchases with nutritional benefits. Which are the food groups more suitable for a price change ? Where are the more disparities in price responsiveness among income classes ? Are own-price effects the only relevant ? Do cross-price effects matter ? To study food demand, we estimate an Exact Affine Stone Index (EASI) demand system developed by Lewbel and Pendakur (2009) and recently implemented by Zhen et al. (2014) for beverage and food demand. This specification is more flexible than the popular Almost Ideal Demand System (AIDS) of Deaton and Muellbauer (1980) and the subsequent literature. For 21 food groups built according to their environmental and nutritional characteristics, we run expenditure and price elasticities. We obtain these results on four income classes and four age groups in order to assess socioeconomic and life-cycle inequalities in demand. The data cover the period 1998-2010. Our paper involves several contributions to the study of food demand in a perspective of simulating health and environmental-friendly policies. First, we use the utility-theoretic EASI demand system to characterize household purchases on food and beverage preferences. To our knowledge, this is the first application on French data. Second, we develop this approach on 4 a large panel dataset, an estimation framework which was not addressed till now. Third, we study the implications of social differentiation of food patterns on public policies by taking into account income inequalities and life-cycle effects. It provides detailed results on the eventuality of various patterns of diet substitution and price responsiveness. This article is organized as follows. Section 2 describes the EASI demand system and section 3 the data and the methodology implemented. Section 4 presents the estimation results and comments the implications for policies. Section 5 concludes. 2 Specification and econometric consumption model We retain the EASI demand system developed by Lewbel and Pendakur (2009) to describe food demand functions. This approach uses an utility-derived model and non linear Engel curves thus giving more flexibility to the demand specification. It enables to measure socioeconomic inequalities and life-cycle effects in food consumption. Following Zhen et al. (2014), we consider an incomplete demand system to model food-at-home purchases which implies a strong assumption of weak separability. In particular, this implies that we use food expenditure instead of income to design consumer demand (Blundell and Robin (1999)). The EASI demand system share with the AIDS some desirable properties. In particular, it is linear and enables to aggregate over consumer preferences. It also has several advantages because it defines implicit Marshallian demand functions with flexible Engel curves. These demand functions are given by defining the implicit utility as the log of food expenditure deflated by the log Stone price index. Therefore, this specification uses an exact deflator, and not an approximated expenditure. Here, the EASI demand system is based on cohort observations c1 . Each cohort budget1 This is justified by the structure of our data, see section 3. 5 shares for food group i(i = 1, ..., N ) is defined as the sum over households’ budget-shares: wict = 1 Nct P wiht , for h (h = 1, ..., H) households, c = 1, ..., C cohorts and t(t = 1, ..., T ) time periods; wiht is the household h budget-share for food group i and Nct is the number of households within cell c in t. The demand for each food group is defined as a function of prices, food expenditure and socio-demographic characteristics: (1) wict = R X r βir (yct ) + r=1 L X l=0 αil Zl,ct + N X γi ln pict + uict , i=1 where pict is the price for food group i, cohort c at period t ; yct is an implicit utility level at the cohort level and for each time period t. Its polynomial degree r enables the flexibility of Engel curves; Zl,ct include l(l = 1, ..., L) socio-demographic characteristics; β, α and γ are the parameters to estimate; and uict is the residual. More precisely, the implicit utility level is given by: (2) yct = ln(xct ) − N X i=1 N 1X wict ln pict + γij ln pict ln pjct . 2 i,j=1 Plugging Equation (2) into (1) gives a demand system which is not restricted by Gorman rank conditions. It can be see that Engel curves depend on each food group through β parameters, which illustrate the shape of the Engel curve. The polynomial degree r is empirically chosen to fit the data thus giving flexibility to the demand system. This specification enables to exploit the unobserved heterogeneity through the error term. Note that because w appears on both sides of the demand equations, controlling for endogeneity enables to obtain efficient results. 6 3 Data 3.1 Data description We use Kantar Worldpanel data from 1998 to 2010. Each annual survey contains weekly food acquisition data for an average of 15,000 households, with an annual rotation of one third of the participants. The households are selected by stratification according to several socioeconomic variables, and remain in the survey for a mean period of four years. All participating households register the grocery purchases through the use of bar codes. From 1998 to 2008, to register grocery purchases without a bar code, each household is assigned to one of two groups to alleviate its workload. Each group (half of the survey) is requested to register its purchases for a restricted set of products: meat, fish, and wine for the first group, and fresh fruits and vegetables for the second group. For 2009 and 2010, the two sub-groups are associated into a single group. Hence, the survey gives the food purchases for more than 8,000 households for 169 periods of four weeks spanning over 1998 to 2010. We grouped food items into 21 categories taking into account the environmental emissions and the nutritional contents of the products (according to Masset et al. (2014) results), consumer preferences and consumer willingness to substitute products within categories of foods, see Table 6. [Insert Table 6 here.] Nutrients characteristics are those of Vieux et al. (2013), and concern more than 500 food products. For each product, the selected nutrients presented here are energy intake (measured in food calories), and proteins, plant-based proteins, animal proteins, saturated fats, cholesterol, vitamin B12, vitamin D, iron and sodium. Environmental data2 are collected by Greenext, an environment consultancy, to assign the 2 Version of the data set 06/11/2013. 7 environmental impact of products through Life Cycle Analysis. The data set delivers, for 311 products, the environmental impact of producing these products. They are illustrated by the following three variables : (1) CO2 relates to the impact on climate change (in gram of CO2 per 100g); (2) SO2 relates to air acidification (in gram of SO2 per 100g); (3) N relates to Nitrates outputs (in gram of N per 100g). 3.2 Cohort Construction We define 48 cohorts to capture both income effects, life-cycle effects, and regional heterogeneity. They are constructed on the following variables: 1. Four income classes, based on family income corrected by consumption units according to OECD scale (Gardes et al. (2005); Allais, Bertail, and Nichèle (2010)): modest, lower average, upper average, well-off; 2. Four age classes based on the age of household head: under 30 years old, 31-45, 46-60, over 61; 3. Three regions with significant differences over food groups consumptions and expenditure: Paris and its suburbs; the North and East; the South and the West3 . Hence household data are aggregated to obtain a pseudo panel and recover the total foodat-home expenditure (Allais, Bertail, and Nichèle (2010)). Therefore, our data do not include infrequency of purchase issue, contrary to Tiffin and Arnoult (2010). Here we consider that the absence of purchase during a time period is a true zero corresponding to non consumption. Hence, this dataset is made by 48 cohorts and 169 time periods, i.e. 8112 observations. The descriptive statistics of this sample are presented in Table 7. 3 Variation between regions is very low (see Allen (2010)) hence with three regions we capture the main differences on food consumption. 8 [Insert Table 7 here.] At the cohort level, and for each time period, we compute for the food purchases: the total emissions (in terms of SO2 , CO2 and N) and the nutrient intakes, see column (a) of Table 8. Based on the food purchase basket, these values enable to compute the contribution of each food group to the full emissions (resp. nutrient intakes). [Insert Table 8 here.] 3.3 Descriptive Statistics The set of sociodemographic variables available in our data allows to characterize the modest households (lowest income group) as those living with income per unit of consumption inferior to 1500e/month, whose reference person has an education level inferior or equal to baccalaureate, mainly blue collars or retired. The majority of these households include children under 16 years. The level of equipment is inferior compared to other income groups: on average, these households have at home less computers, or dishwashers. These characteristics induce budget and food purchases disparities, which can be translated in terms of nutritional content and environmental impact. Socioeconomic disparities in food purchases, nutritional content and environmental impact At the global level, we observe a structure of purchases where food products including animal content represent the major part of the budget (55.6%). This structure varies with age and income classes. Among youngest households (with a head of less than 30 years), animal products represent a lesser share in modest (49.6%) than in well-off households (52.6%)4 . Concerning the nutritional content, note that those disparities vanish: the share of animal source food products 4 Detailed tables per income and age classes are available upon request. 9 in total energy remains around 45.5%. Similarly, the share of animal proteins in total proteins does not vary with income class, around 76.5%. Among the 21 food and beverage groups, plantbased fats and yogurts are the two largest sources of energy. This remains true for the two lowest income classes, while cheese overpasses yogurts in both higher income classes (well-off and upper-average households). Concerning the environmental impact of purchases, our computation amounts to 1.43 tons CO2 eq. or 3.9kg per household. Figure 1 shows that animal products account only for half of these emissions, near the half for nitrates but 3/4 of SO2 impact. Highest CO2 emissions come from modest households (3.9kg) compared to well-off ones (3.4kg). Combining with age, we observe higher emissions with age 45-60, the least among upper-average and well-off households of less than 30 years. Quite similarly, highest SO2 emissions come from modest households with age 30-45 and 45-60 and the least among upper-average and well-off households of less than 30 years. For nitrate emissions, the results are only slightly different: highest values are observed for modest households with age 30-45 and 45-60, and the least for well-off households less than 30 and 30-45 years. In comparison for CO2 emissions, Chancel (2014)’s work based on French overall consumption found significant life-cycle effects, since older cohorts were the more emitting. Income effects showed that richer were emitting more than poorer. This is in tune with the range of disparities observed in the calorie and protein contents. The larger environmental emissions and calories content found in modest households purchases correspond to different food patterns, in particular between food-at-home and food-away-fromhome. In a previous analysis of French budget data, Caillavet, Lecogne, and Nichèle (2009) found that the budget share for food consumed at home is higher for modest households. Moreover, the demographic composition of the households may differ between the different income 10 and age classes. 4 Estimations and Results Before to describe the estimation of the demand system defined in Section 2, this section presents the prices construction. 4.1 Recovering prices from unit values Given we do not observe the price paid by households within the data, they are approximated by using food groups expenditures and quantities purchased. This gives unit values instead of real prices. Indeed, demand estimates are very sensitive to measurement units (Moschini (1995)), therefore we follow Crawford, Laisney, and Preston (2003) to compute them from a first stage estimation. There are several reasons for this. First, the structure of our data does not enable to follow Cox and Wohlgenant (1986) and Park and Capps (1997). Indeed, their methodology imposes to observe households total food-at-home expenditure and to regress in a first step unit values on a total food-at-home expenditure and some socio-demographics variables. Then their estimation results are used to control for the unobserved heterogeneity among prices. In particular, regressors introduce the sum of the intercept and the residual of the first stage regression. Second, without observations on the price paid by households, Deaton (1988) suggests to approximate the actual prices by unit values. He used clustered observations thus controlling for the unobserved geographical heterogeneity among observations. Here, we construct unit values from the total food at-home expenditure and quantities of 11 food group at the cohort level by: ln ψict = ηc0 + (3) K X ηik Z∗k,ct + ζict ln qict + δict ln xct + ict , k=1 where ψict is the unit value for food group i, cohort c and period t; Z∗k,ct stands for the sociodemographic variable k; qict is the quantity of i purchased; xct is the total food expenditure of cohort c in t; η, ζ, and δ are the unit value equation parameters to be estimated; ict is the residual. Adjusted unit values (lnd ψict ) from this equation are used as proxy for prices (ln pict ) into the demand system. Accordingly, we have a vector of c prices per period corresponding to one price per cohort with each unit values corrected from quality effects (Beatty (2010)). Unit value regressions are made for each food group at the cohort level by regressing unit values on the log-food-at-home expenditure (ln(x)), log-quantities (ln(q)), income classes, time period dummies5 , and three variables illustrating households durable ownership. These latter variables are fryer, dish washer, and freezer6 . As for the demand system, they are constructed by cohort and indicate the proportion of each household owning these equipments. Estimation results are presented in Table 9. We use a Fixed Effects (FE) estimator, which produce unbiased and consistent parameters estimates with T = 169. We find that quantity parameters are all negative and significant at the level of 5%7 . For example, a 1 % increase of quantity of juices reduces their demand by 0.10%. Food expenditure coefficients are always positive and significant. For juices, alcohol, bottled water, beef, other meats and cheese, our estimates indicates that a 1% increase of food-at-home expenditure increases the unit value of this food group by more than 1%, ceteris paribus. [Insert Table 9 here.] 5 We have 169 time periods, which are constructed from 13 periods per years of 4-weeks purchases. Therefore, we introduce 13 years and 12 periods of 4-weeks dummies. 6 Note that these variables are not included into the demand system. 7 This level is retained for all the comments in this article. 12 4.2 Demand estimates and elasticities The demand system includes each food group unit values, food-at-home expenditure, and the socio-demographic variables presented in Table 7. The demand system is estimated without the last equation (for prepared desserts) whose parameters are recovered by using the theoretical restrictions (symmetry, homogeneity and adding-up).We demeaned variables to estimate a SURE demand system. For each value of r (starting from 1 to 6), we run 2SLS estimator. Based on Wald tests, we retain r = 4. Hence, 72% of the polynomial degree coefficients are significant. Lewbel and Pendakur (2009) and Zhen et al. (2014) retained r = 5 based on Wald test for the last degree. Estimation results are available upon request, and are used to compute several kind of elasticities. Food Group Elasticities Expenditure, compensated own and cross-price elasticities of food groups are presented in Table 1. They are computed for the sample median of each variable. To measure socioeconomic inequalities, we compute them at the median value of each income and life-cycle classes, see Tables 2 and 3. [Insert Tables 1 to 3 here.] To measure the impact of price variations on household purchases, first, we compute the food expenditure elasticities (eEX i ) as follows: 0 eEX = (diag(W ))−1 [(IJ + BP )−1 B] + 1J , i where W is a vector of budget shares, B = PR r=1 β̃ict,r y rct , P = ln pict . The elasticities are conform with economic theory in the sense that we find positive food expenditure elasticities. At the global level, we observe higher elasticities (slightly over 1) for 13 soft drinks, coffee and tea, and plant-based foods high in fats. Disaggregated by income and age classes, food expenditure elasticities do not show much variation, see Table 3. For the modest households, alcohol and meats are the more budget sensitive, so as for well-off households. There is more variability among food groups than among income and age classes. Among drinks, soft drinks and bottled water elasticities show some variation with income and life-cycle classes. For example, an increasing budget means a higher increase of soft drinks for older households (reference person over 60 years) and richer ones. Juices elasticities are higher for modest households, in particular over 60 years. Among other foods, plant-based fats and fish elasticities are higher with increasing age. Conversely, starchy foods, processed fruits and vegetables elasticities decrease for all age groups above 30 years. Second, we compute two kinds of price elasticities. Indeed, the EASI is directly derived from a cost function, therefore, it estimates Hicksian demand functions. These latter enables to C ) (or Hicksian price elasticities). They are given compute compensated price elasticities (eEP ij by: C = −δij + eEP ij γij + wj , where δij = 1 if i = j. wi These elasticities give the impact of 1% price increase at a constant utility level. However, to design policy issues, uncompensated price elasticities are commonly used because they enable to consider a constant food expenditure in consumer choices. Hence these elasticities are computed C with eEX and eEP by: i ij NC C eEP = eEP − wj eEX ij ij i . As expected, compensated own-price elasticities are negative and all significant, see Table 14 1. Their values are in the same range than Allais, Bertail, and Nichèle (2010)’s results, which were obtained with an AIDS estimation. Beverages show the highest price sensitivity, in particular soft drinks. Note that alcohol price elasticity decreases with age and income, while soft drinks price elasticity increase strongly in both dimensions (reaching -1.69 for well-off households whose head is over 60). Contrary to other age groups, juices show a higher elasticity than alcohol above 45 years. Soft drinks and juices are the products with more price sensitivity variations according to income classes at the extreme age groups (younger and older households). Excluding beverages, higher values are observed for products of animal origin: animal fats, meats excluding beef (i.e.mainly poultry and pork), and yogurts. These foods groups are specially price sensitive. Cooked meats, fish and seafood, and beef price elasticities remain are around 1.2. Yogurts elasticities vary with income : they are lower for modest households and increasing over 45 years, in particular for well-off households. Animal fats and meats other than beef have a price sensitivity driven by income more than by age of the household head. Controlled by age groups, we observe a varying price sensitivity to income for yogurts, animal fats, and other meats (poultry, pork). Here, well-off households show higher elasticities compared to modest households. Among plant-based products, processed fruits and vegetables and starchy foods are the more price sensitive. Plant-based products show less varying elasticities: flat with income, except starchy foods which are more price sensitive for well-off households in the 45-60 age group. Plant-based fats elasticities are higher for modest households under the age of 45, and lower after 60. Note that fresh fruits and vegetables price sensitivity is the same among income classes (but increasing with age of the household head till 60 years). Consequently, when restricting to own-price effects, food groups more suitable for intervention in terms of price sensitivity, 15 are firstly beverages: soft drinks, juices, and bottled water; then animal products: yogurts, animal fats and other meats; among plant-based products: processed fruits and vegetables, starchy foods, and plant-based fats. However overall food purchases, energy intake and environmental impacts also depend on the signs and magnitude of cross-price elasticities between the 21 foods and beverages groups, and in particular between animal and plant-based products. They are presented in Table 1. They all have low values since 0.15 is the maximum. This result differs in particular from the high values obtained by Zhen et al. (2014) in an EASI setting, but based on individual data and not cohorts which have a smoothing effect due to loss of variability. In our estimation, compared to own-price elasticities values, cross-price elasticities could in most cases be considered as negligible. We comment in the following the relationships reaching at least 0.1. Those relationships are substitutions, since all complementarities have low values except in the case of mixed-origin prepared dishes. The cross-price elasticities observed (see Table 1) , when significant, show us both kinds of substitutions: between animal and plant-based products, and within each same origin of products. They also implicate beverages with other food products, showing that liquid and solid part of the diet interfere strongly. Indeed, focusing on animal products, we find that beef purchases substitute mainly with cooked meats, but also with plant-based foods high in sugar and prepared dishes. We observe the same relationships with cooked meats but with a lesser magnitude. Conversely, the group of other meats (mainly pork and poultry) substitute with plant-based foods: fruits and vegetables products (fresh, processed, juices), starchy foods, mixed-origin prepared dishes. Fish and seafoods seem to be a special case: they show higher values for substitutions: with plant-based foods high in sugar (i.e. biscuits and confectionary) and with plant-based dishes. 16 They substitute also with several beverages: alcohol and coffee and tea. Animal fats do not show substitutions nor complementarities with noticeable values. Concerning cheese purchases, we find substitutions with mixed-origin prepared dishes and more unexpectedly with fresh fruits and vegetables, bottled water and soft drinks. Note that Zhen et al. (2014) found also substitution between cheese and soft drinks. Yogurts purchases substitute with plant-based fats and juices. Finally, origin-mixed prepared dishes interact with a large number of other food groups. They substitute mainly with plant-based foods high in sugar, fresh fruits and vegetables, cheese, soft drinks, other meats. Turning to income and life-cycle effects differences in price responsiveness, we find evidence of limited differentiation for some food groups. Beverages are driven by age more than income (older households showing more price sensitivity to juices and lower for soft drinks, bottled water and alcohol). Conversely to previous estimations on French data (Bertail and Caillavet (2008), Allais, Bertail, and Nichèle (2010)), fruits and vegetables -processed or fresh- have stable price sensitivity according to the income class, but older households have a higher price sensitivity to processed ones. We find as well that disparities in animal or plant-based foods are more driven by age than by income at this stage (for example animal fats and other meats while cooked meats and beef are neutral). Several reasons may explain the weakness of these results. First, it is known that the level of aggregation of data in food groups (21 in our case) does not allow to detect differentiation of purchases through brands and quality within a same food group, which is probably at this level the main strategy of low income households (Beatty (2010)). Second, the aggregation of data through cohorts, implied by the structure of Kantar data, induces mechanically a loss of variability. Finally, we deal with purchases at the level of the household. Therefore the variable of age of the household head becomes a proxy for household structure (determined by the position in the life-cycle : number of members, presence of children) which is a major 17 determinant of the composition of food purchases. Indeed, the two medium age groups (30-45 and 46-60 years), corresponding to working-age adults, show very similar elasticities. The age variable probably captures most part of the remaining variability left by the cohort structure of the sample. NC , we compute environmental and Nutrients and environmental elasticities Finally, with eEP ij nutrients elasticities (Huang (1996); Huang and Lin (2000); Allais, Bertail, and Nichèle (2010)). For each indicator `, they are given by: Ind` = S` × D` , (4) where Ind` denotes the matrix of nutrient and environmental price elasticities, S` is a matrix including the food share of indicator `, and D` is a matrix of price elasticities. Hence, we measure the impact of 1% increase of prices on the amount of environmental emissions and nutrient intakes. Such elasticities enable to compute the impact of a price increase on some targeted food products on environmental emissions and nutrient intakes. Nutrient response to price changes are presented in Table 4. They have very low values. This result confirms previous literature (Huang and Lin (2000); Beatty and LaFrance (2005) ; Allais, Bertail, and Nichèle (2010)). [Insert Table 4 here.] Sociodemographic effects Education does not introduce different results (see Table 5): less educated (less than baccalaureate) or more educated (more than baccalaureate) have the same sign and range of coefficients for animal products. The presence of children under 15 years is more relevant: it favours the budget share dedicated to beef (at the level of significativity of 10%8 ) 8 For this section the level of 10% is retained because coefficients are not significant at the level of 5 %. 18 and other meats, prepared dishes of mixed origin, yogurts at the expenses of cooked meats and animal fats. Farmers favour animal fats and cheese, and fish. Executives favour cooked meats, animal fats, cheese, fish and yogurts. [Insert Table 5 here.] 5 Impact of taxation : alternative scenarios Finally, we implement two scenarios of tax, which are equivalent to a VAT increase: • In the environment-friendly scenario, namely ENV, we increase by 20% the prices of all products with animal contents. The tax concerns beef, other meats, cooked meats, fats from animal production, cheese, fish and seafoods, yogurts, and prepared mixed meals. These goods have a high environmental impact, see Masset et al. (2014). • In the environment and nutrition-friendly scenario, namely ENV-NUT, we increase by 20% the prices of food categories most adverse to both environment and health. The tax is implemented on beef, cooked meats, fats from animal production, cheese, and prepared mixed-origin dishes. The results are presented in columns (b) of Table 8. To assess the attractiveness of our two scenarios, our focus is primarily on the reductions obtained in the environmental impacts through SO2 , CO2 and N emissions. Then we consider the induced changes in dietary health indicators, such as the changes in total calories and in related animal sources of nutrients, which can be positive or negative for health. At the environmental level, columns (b) of Table 8, the scenario of taxing all animal products with a 20% rate, ENV, has logically more impact than the mixed scenario, ENV-NUT. ENV 19 decreases emissions from 10% (CO2 and N) to 16.6% (SO2 ). Compared to ENV, ENV-NUT displays inferior reductions in emissions, in particular for N (-30%). Concerning CO2 , total yearly reduction would reach 117kg (ENV-NUT) to 142kg (ENV) CO2 equivalent per person9 . This is comparable to the Danish estimates using consumer data which finds a 112kg to 277kg reduction when taxing all foods according to their polluting potential (Edjabou and Smed (2013)). No other literature were found to compare interventions on levels of SO2 and N emissions. At the nutritional level, reductions in total energy content from purchases vary from -8.8% in ENV to -5.4% in ENV-NUT. In ENV, interesting reductions concern animal proteins (-21.5%) and total proteins (-17%). Vegetal proteins also decrease slightly due to the taxation of products mixing animal and vegetal contents (prepared mixed meals and prepared desserts). Health favorable effects include decreases in saturated fats (-15.6%), cholesterol (-18.7%) or sodium (-12.5%). However, we observe undesirable effects such as a strong decrease in key nutrients: vitamin B12 (-20.6%), vitamin D (-18.8%) and iron (-6.4%). In ENV-NUT, total and animal proteins reductions are only half of those observed in ENV. This leads to lower effects on unhealthy nutrients such as sodium or saturated fats, but also on losses of good nutrients. In terms of environmental/nutritional compatibility, ENV-NUT appears more balanced: the environmental impact on CO2 is not so different from ENV, saturated fats and sodium reductions are still consistent, and good nutrients (here vitamins B12 and D, and iron) losses are more limited than in ENV. We find that taking into account health constraints is not so costly since it reduces CO2 by 17.7%. Income and life-cycle effects For a given income class, we can observe life-cycle effects, see Table 10. In both scenarios, regarding emissions, age introduces more variations in the well9 Computations are based on daily emissions of food purchases. Details are available upon request. 20 off income class, especially in N emissions, but this effect remains very moderate. From a nutritional perspective, there are more differentiated effects: reductions in proteins are observed mainly among the 30-45 years, while reductions in saturated fats are more important among the 60 years and over. [Insert Table 10 here.] For a given age group, we observe income class effects. Here, variations in emissions are a little higher since the rate of change induced by taxation emissions decline from modest to well-off households. However, the nutritional content of the food basket does not appear to vary much with income class. Though moderate, these effects suggest that taxation would induce more differentiated impacts on emissions according to income class, while nutritional changes would induce more disparities among age groups. Combining income and age effects, in both scenarios the highest rate of reduction in environmental impacts is observed for the young (less than 30 years) and lower-average income households (in ENV, -17% for SO2 and -10% for CO2 and N). However in levels, higher total yearly impacts are obtained for households which are in the middle of the life-cycle. Among the various environmental indicators we use, the greatest impact is on SO2 emissions, which decrease (around -16%), corresponding to the highest reduction in animal proteins (approximately -22%). More precisely focusing on CO2 in the ENV scenario, the highest rate of CO2 reduction is observed for modest or lower-average income households whose head is under 30. In that case, yearly reductions amount to 160kg10 . However, with a smallest impact rate, total yearly reductions reaches 179kg for the modest households which are in the middle of the life-cycle 10 Computations are based on daily emissions for each income and life-cycle classes of food purchases. Details are available upon request. 21 (household head between 45 and 60 years). The smallest effect accounts for the young and welloff households which has a lower CO2 content of purchases and therefore reaches only 100kg reduction. Similar impacts can be calculated concerning SO2 and N emissions. These disparities are partly explained by different patterns of food-at-home consumption, probably more than by variations in price responsiveness as shown by the elasticities. Trade-offs between the two scenarios are however larger for the older households, be they well-off or modest (-18.1% loss in CO2 reduction) than for the younger ones (-16.8%). 6 Conclusions In a sustainability perspective, this article examined the impact of French food-at-home purchases on both environmental emissions and nutritional intakes. Furthermore, two alternative price policy scenarios are explored in an attempt to combine environmental and health objectives. For this, we built 21 food groups distinguishing plant-based and animal-based products. We computed food demand elasticities from the EASI demand system estimated on a sample covering 13 years (1998-2010). This demand system allows for flexible Engel curves for each food group. We found evidence of the relevance of this model since the best fit of our estimation was obtained for a polynomial specification of range 4. On this basis, we computed the mostly used CO2 emissions elasticities, and also SO2 and nitrate ones. This is an attempt to widen the range of indicators to take into account different aspects of pollution. To assess possible health effects, we ran as well nutrient elasticities. Finally, we computed disaggregated elasticities over income and life cycle-effect classes in order to take into account the social differentiation of food patterns and price responsiveness. At this point of research, the price elasticities of emissions and nutrients do not induce very important income 22 disparities, but our results on life-cycle effects are noticeable. We are thus able to address key issues in food policy design. Are animal products sensitive to a price policy and possible to target ? Among food groups, soft drinks, animal fats, yogurts and meats other than beef show the highest responsiveness to prices. At the same time, we find high values of environmental and nutrient elasticities for meats and soft drinks. Our results suggest that taxing these food groups may be a tool for an environmental-friendly policy. This is supported by the fact that we find many interactions in terms of substitutions between all types of foods, be they liquid, solid, plant-based or animalbased. However, their range remains quite limited compared to own-price values. Is the environmental impact very sensitive to the use of different indicators ? Our results evidence that the CO2 emissions, which is the indicator most used in the literature, does not show the strongest results. We find that the price sensitivity of SO2 emissions is higher. Therefore, widening the range of indicators studied could be crucial for the conclusions and is a useful input of our study. What would be the health implications of an environmental tax designed on animal products? Focusing on eight groups of animal products, which appeared the best candidates to tax on environment grounds, we could appreciate that the nutritional impact is far from neutral. Strong favourable effects on fats or cholesterol decreases coincide with undesirable ones, in particular reduced intakes of key micronutrients such as vitamins and iron. A second scenario which takes into account also health considerations and applies a price increase to five animal products groups only, does not yield the same reduction in emissions but limits the unfavourable nutritional effects. Indeed, in the scenarios presented here, an improvement in health impact would lead to 18% loss of CO2 reduction. 23 In conclusion, our study illustrates in the French case the complexity of food demand through the multiple relationships observed between animal and plant-based food groups. It also shows the ambiguities of combining environment with nutrition objectives, through the choice of the food groups on which a price increase should be applied. Obviously, a trade-off cannot be avoided between both constraints. Our analysis shows that it may be not so costly. Of course, more economic scenarios should be implemented. For this, a combined set of healthy and environmental-friendly guidelines from nutritional experts is certainly one issue deserving other developments in the future. Moreover, this perspective may be improved by computations of Consumer Surplus to measure these different public policies. 24 Bibliography Allais, O., P. Bertail, and V. Nichèle. 2010. “The effects of a fat tax on French households’ purchases: a nutritional approach.” American Journal of Agricultural Economics 92:228–245. Allen, T. 2010. “Impacts des variations de prix sur la qualité nutritionnelle du panier alimentaire des ménages français.” PhD dissertation, Université Montpellier 1. Aston, L.M., J.N. Smith, and J.W. Powles. 2012. “Impact of a reduced red and processed meat dietary pattern on disease risks and greenhouse gas emissions in the UK: a modelling study.” BMJ open 2. Beatty, T.K. 2010. “Do the poor pay more for food? Evidence from the United Kingdom.” American Journal of Agricultural Economics 92:608–621. Beatty, T.K., and J.T. LaFrance. 2005. “United States demand for food and nutrition in the twentieth century.” American Journal of Agricultural Economics 87:1159–1166. Bertail, P., and F. Caillavet. 2008. “Fruit and vegetable consumption patterns: a segmentation approach.” American Journal of Agricultural Economics 90:827–842. Blundell, R., and J.M. Robin. 1999. “Estimation in Large and Dissaggregated Demand Systems: An Estimator for Conditionally Linear Systems.” Journal of Applied Econometrics 14:209–232. Boeglin, N., C. Bour, and M. David. 2012. “Le contenu carbone du panier de consommation courante.” CGDDSOeS, Observation et Statistiques Environnement 121. Briggs, A.D., A. Kehlbacher, R. Tiffin, T. Garnett, M. Rayner, and P. Scarborough. 2013. “Assessing the impact on chronic disease of incorporating the societal cost of greenhouse gases into the price of food: an econometric and comparative risk assessment modelling study.” BMJ open 3:e003543. Caillavet, F., C. Lecogne, and V. Nichèle. 2009. “La fracture alimentaire : des inégalités persistantes mais qui se réduisent.” La Consommation, INSEE Références, pp. 49–62. Capacci, S., M. Mazzocchi, B. Shankar, J. Brambila Macias, W. Verbeke, F.J. Pérez-Cueto, A. Kozioł-Kozakowska, B. Piórecka, B. Niedzwiedzka, D. D’Addesa, et al. 2012. “Policies to promote healthy eating in Europe: a structured review of policies and their effectiveness.” Nutrition reviews 70:188–200. Chancel, L. 2014. “Are younger generations higher carbon emitters than their elders?: Inequalities, generations and CO2 emissions in France and in the USA.” Ecological Economics 100:195 – 207. Cox, T.L., and M.K. Wohlgenant. 1986. “Prices and quality effects in cross-sectional demand analysis.” American Journal of Agricultural Economics 68:908–919. Crawford, I., F. Laisney, and I. Preston. 2003. “Estimation of household demand systems with theoretically compatible Engel curves and unit value specifications.” Journal of Econometrics 114:221–241. Deaton, A. 1988. “Quality, quantity, and spatial variation of price.” The American Economic Review, pp. 418–430. Deaton, A., and J. Muellbauer. 1980. “An almost ideal demand system.” The American Economic Review 70:312– 326. Drewnowski, A., E.C. Fiddler, L. Dauchet, P. Galan, and S. Hercberg. 2009. “Diet quality measures and cardiovascular risk factors in France: applying the Healthy Eating Index to the SU. VI. MAX study.” Journal of the American College of Nutrition 28:22–29. Edjabou, L.D., and S. Smed. 2013. “The effect of using consumption taxes on foods to promote climate friendly diets–The case of Denmark.” Food Policy 39:84–96. 25 Gardes, F., G.J. Duncan, P. Gaubert, M. Gurgand, and C. Starzec. 2005. “Panel and pseudo-panel estimation of cross-sectional and time series elasticities of food consumption: the case of US and polish data.” Journal of Business & Economic Statistics 23:242–253. Hébel, P., and F. Recours. 2007. “Effets d’âge et de génération: transformation du modèle alimentaire.” Cahiers de nutrition et de diététique 42:297–303. Hoolohan, C., M. Berners-Lee, J. McKinstry-West, and C. Hewitt. 2013. “Mitigating the greenhouse gas emissions embodied in food through realistic consumer choices.” Energy Policy 63:1065–1074. Huang, K.S. 1996. “Nutrient elasticities in a complete food demand system.” American Journal of Agricultural Economics 78:21–29. Huang, K.S., and B.H. Lin. 2000. Estimation of food demand and nutrient elasticities from household survey data. 1887, US Department of Agriculture, Economic Research Service. Lewbel, A., and K. Pendakur. 2009. “Tricks with Hicks: The EASI demand system.” The American Economic Review, pp. 827–863. Mackenbach, J.P., I. Stirbu, A.J.R. Roskam, M.M. Schaap, G. Menvielle, M. Leinsalu, and A.E. Kunst. 2008. “Socioeconomic inequalities in health in 22 European countries.” New England Journal of Medicine 358:2468–2481. Masset, G., L.G. Soler, F. Vieux, and N. Darmon. 2014. “Identifying Sustainable Foods: The Relationship between Environmental Impact, Nutritional Quality, and Prices of Foods Representative of the French Diet.” Journal of the Academy of Nutrition and Dietetics, pp. . McMichael, A.J., J.W. Powles, C.D. Butler, and R. Uauy. 2007. “Food, livestock production, energy, climate change, and health.” The Lancet 370:1253–1263. Moschini, G. 1995. “Units of measurement and the Stone index in demand system estimation.” American Journal of Agricultural Economics 77:63–68. Ovrum, A., G.W. Gustavsen, and K. Rickertsen. 2014. “Age and socioeconomic inequalities in health: Examining the role of lifestyle choices.” Advances in Life Course Research 19:1–13. Park, J.L., and O. Capps. 1997. “Demand for prepared meals by US households.” American Journal of Agricultural Economics 79:814–824. Scarborough, P., S. Allender, D. Clarke, K. Wickramasinghe, and M. Rayner. 2012. “Modelling the health impact of environmentally sustainable dietary scenarios in the UK.” European journal of clinical nutrition 66:710–715. Smed, S., J.D. Jensen, and S. Denver. 2007. “Socio-economic characteristics and the effect of taxation as a health policy instrument.” Food Policy 32:624–639. Tiffin, R., and M. Arnoult. 2010. “The demand for a healthy diet: estimating the almost ideal demand system with infrequency of purchase.” European Review of Agricultural Economics 37:501–521. Vieux, F., L.G. Soler, D. Touazi, and N. Darmon. 2013. “High nutritional quality is not associated with low greenhouse gas emissions in self-selected diets of French adults.” The American journal of clinical nutrition 97:569– 583. Westhoek, H., J.P. Lesschen, T. Rood, S. Wagner, A. De Marco, D. Murphy-Bokern, A. Leip, H. van Grinsven, M.A. Sutton, and O. Oenema. 2014. “Food choices, health and environment: effects of cutting Europe’s meat and dairy intake.” Global Environmental Change, pp. . Zhen, C., E.A. Finkelstein, J.M. Nonnemaker, S.A. Karns, and J.E. Todd. 2014. “Predicting the Effects of SugarSweetened Beverage Taxes on Food and Beverage Demand in a Large Demand System.” American Journal of Agricultural Economics 96:1–25. 26 27 Juic Alc Soda Wat Cof FF&V Grains VHF VD VHS Starch PF&V Beef OM CM AHF Cheese Fish Yogurt PrepM Food X Juic -1.295 0.091 0.229 0.056 0.098 0.011 0.093 0.185 0.067 -0.002 0.068 0.087 0.091 0.110 0.093 0.161 -0.102 0.020 0.119 0.005 0.988 Alc 0.176 -1.206 0.300 0.072 0.149 0.025 0.104 0.116 0.060 0.274 0.081 0.166 0.050 0.114 0.146 0.120 0.156 0.221 0.116 0.102 0.957 Soda 0.173 0.117 -1.518 0.029 0.080 0.026 0.026 0.045 -0.054 0.013 0.031 0.038 0.033 0.038 0.035 0.073 0.126 0.005 0.056 0.084 1.050 Wat 0.061 0.041 0.041 -1.389 0.030 0.151 0.027 0.073 0.087 0.017 0.146 0.073 0.078 0.015 0.105 0.117 0.227 0.012 0.108 0.092 0.918 Cof 0.090 0.071 0.098 0.025 -0.942 -0.040 0.079 0.039 0.036 0.027 0.023 0.030 0.042 0.017 0.039 0.091 0.003 0.098 0.073 0.054 1.014 FF&V 0.006 0.007 0.018 0.071 -0.022 -1.086 0.011 -0.012 0.021 0.000 0.081 0.031 0.033 0.080 0.041 0.088 0.063 -0.027 0.011 0.070 0.944 Grains 0.028 0.016 0.010 0.007 0.025 0.006 -0.999 0.017 0.013 0.078 -0.004 0.012 0.023 0.025 0.018 -0.006 -0.010 0.028 0.021 -0.030 0.959 VHF 0.099 0.032 0.032 0.036 0.023 -0.013 0.031 -1.171 0.048 0.029 0.124 0.132 0.034 0.040 0.038 0.053 -0.088 0.018 0.095 0.001 1.008 VD 0.051 0.024 -0.055 0.060 0.030 0.032 0.033 0.068 -1.018 0.048 0.044 0.055 0.077 0.033 0.060 0.054 -0.077 0.134 0.062 0.080 0.837 VHS -0.001 0.102 0.013 0.011 0.021 0.001 0.189 0.039 0.045 -1.078 0.049 0.024 0.079 0.043 -0.033 -0.004 -0.035 0.176 -0.018 0.103 0.988 Table 1: Food Expenditure, Compensated Own and Cross-Price Elasticities Starch 0.032 0.020 0.020 0.062 0.012 0.074 -0.006 0.108 0.027 0.032 -1.234 0.064 0.026 0.063 0.053 -0.007 -0.012 -0.004 0.039 0.003 0.848 PF&V 0.039 0.039 0.023 0.030 0.014 0.027 0.018 0.111 0.032 0.015 0.061 -1.242 0.026 0.045 0.040 0.026 -0.025 0.040 0.036 0.001 0.853 Beef 0.155 0.044 0.076 0.122 0.078 0.111 0.131 0.109 0.173 0.188 0.095 0.099 -1.113 0.145 0.183 0.082 0.018 0.108 0.143 0.025 0.960 OM 0.129 0.069 0.059 0.016 0.022 0.182 0.099 0.089 0.051 0.070 0.157 0.118 0.099 -1.340 0.070 0.105 0.088 0.090 0.071 0.123 0.957 CM 0.087 0.071 0.043 0.090 0.039 0.075 0.056 0.066 0.073 -0.042 0.106 0.083 0.100 0.056 -1.199 0.123 0.024 0.001 0.076 0.039 0.976 AHF 0.085 0.033 0.051 0.056 0.052 0.091 -0.010 0.052 0.037 -0.003 -0.008 0.031 0.025 0.047 0.069 -1.402 0.007 0.032 0.025 0.014 0.900 Cheese -0.153 0.122 0.252 0.313 0.005 0.186 -0.051 -0.249 -0.153 -0.074 -0.038 -0.083 0.016 0.114 0.039 0.019 -0.774 -0.007 0.036 0.169 0.374 Fish 0.022 0.124 0.007 0.012 0.116 -0.056 0.101 0.036 0.191 0.264 -0.010 0.097 0.069 0.083 0.001 0.067 -0.005 -1.148 0.085 0.058 0.862 Yogurt 0.141 0.071 0.087 0.117 0.094 0.025 0.083 0.211 0.097 -0.030 0.098 0.094 0.099 0.072 0.096 0.056 0.028 0.093 -1.347 0.035 0.958 PrepM 0.005 0.050 0.107 0.080 0.056 0.130 -0.097 0.002 0.100 0.135 0.007 0.002 0.014 0.100 0.040 0.025 0.106 0.051 0.028 -1.173 0.919 28 Juic Alc -1.237 -1.178 Well-off /-;30] T -20.596 -11.285 -1.239 -1.201 Well-off /]30;45] T -20.792 -12.190 -1.249 -1.204 Well-off /]45;60] T -21.465 -12.297 -1.270 -1.152 Well-off /]60;+ T -22.937 -10.329 -1.334 -1.119 Upper-A. /-;30] T -27.449 -9.188 -1.239 -1.201 Upper-A ]30;45] T -20.800 -12.207 -1.262 -1.223 Upper-A /]45;60] T -22.385 -13.086 -1.243 -1.230 Upper-A /]60;+ T -21.049 -13.395 -1.284 -1.190 Lower-A /-;30] T -23.949 -11.740 -1.360 -1.149 Lower-A /]30;45] T -29.194 -10.218 -1.252 -1.238 Lower-A /]45;60] T -21.667 -13.704 -1.298 -1.259 Lower-A /]60;+ T -24.952 -14.650 -1.273 -1.271 Modest /-;30] T -23.168 -15.161 -1.304 -1.203 Modest /]30;45] T -25.366 -12.251 -1.379 -1.169 Modest /]45;60] T -30.498 -10.934 -1.271 -1.237 Modest /]60;+ T -22.993 -13.680 T-stat are reported below elasticities. Soda -1.511 -37.540 -1.444 -32.413 -1.491 -36.040 -1.522 -38.376 -1.661 -48.514 -1.486 -35.593 -1.417 -30.365 -1.510 -37.485 -1.527 -38.755 -1.616 -45.318 -1.517 -37.982 -1.455 -33.243 -1.513 -37.676 -1.506 -37.185 -1.592 -43.599 -1.525 -38.633 Wat -1.405 -26.063 -1.416 -26.817 -1.405 -26.077 -1.374 -24.061 -1.388 -24.930 -1.415 -26.743 -1.414 -26.649 -1.371 -23.875 -1.356 -22.895 -1.344 -22.129 -1.454 -29.346 -1.417 -26.844 -1.396 -25.453 -1.357 -22.971 -1.350 -22.485 -1.484 -31.421 Cof -0.940 -18.954 -0.943 -21.098 -0.941 -19.418 -0.940 -19.127 -0.941 -19.653 -0.938 -17.989 -0.943 -21.179 -0.943 -20.745 -0.941 -19.446 -0.939 -18.591 -0.941 -19.474 -0.944 -21.724 -0.945 -22.088 -0.942 -20.078 -0.939 -18.427 -0.942 -20.303 FF&V -1.096 -43.516 -1.081 -39.118 -1.082 -39.378 -1.086 -40.574 -1.078 -38.373 -1.093 -42.606 -1.084 -40.087 -1.091 -42.000 -1.087 -40.860 -1.075 -37.369 -1.096 -43.592 -1.091 -42.180 -1.095 -43.109 -1.087 -41.013 -1.078 -38.466 -1.093 -42.605 Grains -0.992 -45.950 -0.993 -46.227 -0.994 -47.194 -0.998 -49.615 -0.999 -50.108 -0.995 -48.033 -0.994 -47.471 -0.997 -49.551 -1.000 -50.282 -1.000 -50.401 -0.998 -49.778 -0.997 -49.414 -1.001 -50.485 -1.001 -50.504 -1.001 -50.504 -1.000 -50.366 VHF -1.139 -33.973 -1.158 -36.792 -1.166 -37.865 -1.178 -39.447 -1.168 -38.132 -1.139 -33.934 -1.170 -38.443 -1.180 -39.746 -1.175 -39.023 -1.154 -36.209 -1.148 -35.376 -1.182 -39.943 -1.193 -41.211 -1.178 -39.415 -1.149 -35.446 -1.146 -35.120 VD -1.007 -23.344 -1.010 -24.188 -1.011 -24.328 -1.021 -27.233 -1.025 -28.761 -1.010 -24.122 -1.009 -23.955 -1.016 -25.896 -1.021 -27.336 -1.025 -28.583 -1.013 -24.913 -1.015 -25.591 -1.019 -26.708 -1.021 -27.502 -1.028 -29.568 -1.012 -24.651 VHS -1.066 -25.336 -1.063 -24.818 -1.072 -26.225 -1.093 -29.098 -1.097 -29.575 -1.063 -24.830 -1.054 -23.389 -1.065 -25.119 -1.089 -28.685 -1.094 -29.298 -1.065 -25.158 -1.043 -21.725 -1.068 -25.619 -1.086 -28.283 -1.093 -29.118 -1.064 -25.081 Starch -1.228 -48.539 -1.220 -47.249 -1.230 -48.903 -1.267 -55.388 -1.273 -56.481 -1.219 -46.989 -1.216 -46.517 -1.229 -48.785 -1.253 -53.080 -1.257 -53.641 -1.214 -46.074 -1.218 -46.790 -1.225 -48.014 -1.243 -51.350 -1.250 -52.457 -1.205 -44.392 Table 2: Compensated Own-price Elasticities by Income and Life-cycle Classes PF&V -1.214 -46.789 -1.207 -45.615 -1.222 -48.357 -1.260 -55.190 -1.261 -55.426 -1.214 -46.903 -1.211 -46.237 -1.227 -49.325 -1.261 -55.381 -1.262 -55.546 -1.217 -47.503 -1.210 -46.165 -1.236 -50.947 -1.260 -55.223 -1.254 -54.190 -1.216 -47.224 Beef -1.114 -12.920 -1.151 -15.093 -1.122 -13.362 -1.109 -12.625 -1.089 -11.562 -1.120 -13.234 -1.146 -14.788 -1.132 -13.954 -1.109 -12.624 -1.093 -11.798 -1.110 -12.666 -1.137 -14.255 -1.123 -13.400 -1.108 -12.573 -1.086 -11.434 -1.110 -12.676 OM -1.363 -23.970 -1.417 -27.530 -1.378 -24.900 -1.353 -23.344 -1.346 -22.859 -1.356 -23.534 -1.408 -26.919 -1.368 -24.278 -1.340 -22.495 -1.332 -21.973 -1.322 -21.361 -1.364 -24.040 -1.335 -22.199 -1.322 -21.338 -1.303 -20.207 -1.307 -20.419 CM -1.217 -27.309 -1.197 -25.120 -1.205 -26.033 -1.216 -27.171 -1.235 -29.243 -1.202 -25.665 -1.194 -24.889 -1.200 -25.464 -1.202 -25.682 -1.221 -27.722 -1.193 -24.725 -1.187 -24.150 -1.182 -23.607 -1.190 -24.434 -1.209 -26.436 -1.182 -23.599 AHF -1.442 -49.646 -1.466 -51.255 -1.465 -51.189 -1.466 -51.243 -1.439 -49.370 -1.400 -46.267 -1.423 -48.196 -1.421 -48.014 -1.411 -47.249 -1.399 -46.218 -1.372 -43.732 -1.389 -45.345 -1.375 -44.068 -1.374 -43.970 -1.345 -41.105 -1.358 -42.418 Cheese -0.757 -11.540 -0.759 -11.401 -0.777 -9.677 -0.783 -8.743 -0.785 -8.425 -0.757 -11.517 -0.768 -10.671 -0.779 -9.442 -0.782 -8.998 -0.783 -8.856 -0.760 -11.327 -0.767 -10.726 -0.780 -9.386 -0.780 -9.366 -0.778 -9.612 -0.758 -11.486 Fish -1.121 -18.297 -1.159 -21.903 -1.143 -20.385 -1.116 -17.921 -1.093 -15.968 -1.133 -19.466 -1.172 -23.119 -1.157 -21.651 -1.137 -19.806 -1.107 -17.090 -1.145 -20.518 -1.186 -24.601 -1.166 -22.540 -1.147 -20.758 -1.105 -16.933 -1.151 -21.147 Yogurt -1.383 -24.454 -1.386 -24.684 -1.385 -24.606 -1.460 -29.401 -1.491 -31.372 -1.342 -21.924 -1.300 -19.452 -1.296 -19.232 -1.388 -24.777 -1.445 -28.392 -1.293 -19.066 -1.227 -15.414 -1.243 -16.279 -1.345 -22.135 -1.416 -26.552 -1.282 -18.405 PrepM -1.169 -23.621 -1.164 -23.124 -1.168 -23.545 -1.174 -24.185 -1.177 -24.488 -1.171 -23.861 -1.154 -22.153 -1.169 -23.613 -1.174 -24.118 -1.201 -26.915 -1.176 -24.333 -1.158 -22.522 -1.167 -23.414 -1.178 -24.541 -1.207 -27.611 -1.171 -23.856 29 Juic Alc Soda Wat Cof FF&V Grains VHF VD VHS Starch PF&V Beef OM CM AHF Cheese Fish Yogurt PrepM LowerAverage /-;30] 0.765 0.963 0.631 0.931 0.901 0.864 0.804 0.874 0.894 0.912 0.855 0.858 0.944 0.927 0.921 0.840 0.835 0.886 0.936 0.918 LowerAverage /]30;45] 0.784 0.966 0.638 0.939 0.918 0.877 0.816 0.881 0.904 0.915 0.865 0.864 0.954 0.940 0.927 0.856 0.879 0.904 0.942 0.925 LowerAverage /]45;60] 0.777 0.971 0.633 0.945 0.922 0.878 0.797 0.879 0.896 0.906 0.851 0.850 0.958 0.945 0.927 0.861 0.893 0.917 0.935 0.926 LowerAverage /]60;+ 0.734 0.973 0.540 0.942 0.918 0.883 0.785 0.882 0.889 0.902 0.846 0.848 0.960 0.945 0.921 0.867 0.897 0.924 0.930 0.924 0.777 0.965 0.687 0.938 0.912 0.876 0.815 0.880 0.905 0.926 0.872 0.871 0.951 0.936 0.930 0.870 0.864 0.895 0.952 0.929 Modest /-;30] 0.807 0.967 0.664 0.948 0.921 0.882 0.811 0.885 0.907 0.927 0.878 0.875 0.957 0.946 0.935 0.881 0.892 0.909 0.957 0.932 Modest /]30;45] 0.785 0.971 0.657 0.950 0.926 0.885 0.799 0.888 0.903 0.914 0.867 0.860 0.961 0.950 0.935 0.884 0.898 0.916 0.947 0.931 Modest /]45;60] 0.738 0.973 0.600 0.950 0.928 0.894 0.790 0.897 0.897 0.910 0.864 0.857 0.963 0.950 0.930 0.885 0.899 0.925 0.940 0.924 Modest /]60;+ UpperAverage /-;30] 0.769 0.964 0.683 0.940 0.914 0.876 0.805 0.880 0.904 0.936 0.878 0.878 0.955 0.945 0.936 0.885 0.871 0.894 0.962 0.931 UpperAverage /]30;45] 0.806 0.967 0.684 0.949 0.922 0.887 0.802 0.887 0.911 0.931 0.888 0.882 0.962 0.954 0.944 0.900 0.901 0.913 0.964 0.937 Table 3: Food Expenditure Elasticities by Income and Life-cycle Classes UpperAverage /]45;60] 0.782 0.970 0.676 0.952 0.926 0.889 0.791 0.890 0.906 0.919 0.876 0.865 0.962 0.953 0.940 0.897 0.898 0.915 0.953 0.932 UpperAverage /]60;+ 0.731 0.972 0.617 0.951 0.930 0.893 0.785 0.901 0.896 0.912 0.870 0.864 0.964 0.954 0.934 0.901 0.892 0.927 0.944 0.923 0.733 0.962 0.636 0.928 0.901 0.874 0.775 0.870 0.900 0.926 0.873 0.868 0.953 0.944 0.931 0.878 0.851 0.885 0.960 0.929 Well-off /;30] 0.773 0.966 0.674 0.945 0.913 0.882 0.760 0.883 0.907 0.927 0.885 0.874 0.961 0.955 0.943 0.902 0.885 0.904 0.966 0.934 Well-off /]30;45] 0.752 0.968 0.664 0.945 0.916 0.882 0.749 0.882 0.896 0.912 0.873 0.854 0.960 0.952 0.936 0.899 0.882 0.902 0.954 0.927 Well-off /]45;60] 0.655 0.967 0.597 0.945 0.916 0.885 0.741 0.890 0.880 0.903 0.852 0.845 0.959 0.950 0.927 0.891 0.871 0.916 0.942 0.917 Well-off /]60;+ 30 Alc Nutrient Elasticities Energy Proteins Animal Proteins Vegetal Proteins Saturated fats Cholesterol Vitamin B12 Vitamin D Iron Sodium -0.057 -0.025 0.000 -0.115 0.000 0.000 0.000 0.000 -0.089 -0.005 -0.071 -0.009 0.000 -0.039 0.000 0.000 -0.021 0.000 -0.080 -0.006 Environmental Emissions Elasticities CO2 -0.104 -0.126 -0.053 -0.066 SO2 N -0.105 -0.074 Juic -0.076 -0.000 0.000 -0.002 0.000 0.000 0.000 0.000 -0.006 -0.013 -0.065 -0.028 -0.050 Soda 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 -0.008 -0.146 -0.046 -0.062 Wat -0.004 -0.019 0.000 -0.086 -0.000 0.000 0.000 0.000 -0.021 -0.005 -0.004 -0.002 -0.004 Cof -0.012 -0.009 -0.000 -0.038 -0.001 -0.001 -0.000 -0.001 -0.021 -0.004 -0.026 -0.012 -0.019 FF&V -0.015 -0.015 0.000 -0.067 -0.004 0.000 0.000 0.000 -0.339 -0.016 -0.026 -0.014 -0.017 Grains -0.187 -0.001 0.000 -0.006 -0.152 -0.000 0.000 0.000 0.000 -0.006 -0.029 -0.011 -0.161 VHF -0.031 -0.014 -0.005 -0.041 -0.026 -0.021 -0.004 -0.019 -0.014 -0.098 -0.019 -0.007 -0.017 VD -0.077 -0.026 -0.009 -0.073 -0.044 -0.012 -0.012 -0.050 -0.038 -0.018 -0.025 -0.024 -0.041 VHS Table 4: Environmental Indicators and Nutrient Elasticities per Food Group -0.073 -0.062 -0.002 -0.277 -0.006 -0.010 -0.002 -0.019 -0.057 -0.042 -0.031 -0.017 -0.027 Starch -0.019 -0.013 -0.002 -0.052 -0.016 -0.012 -0.003 -0.010 -0.030 -0.076 -0.023 -0.009 -0.036 PF&V -0.036 -0.149 -0.187 -0.028 -0.039 -0.082 -0.152 -0.002 -0.089 -0.015 -0.240 -0.437 -0.177 Beef -0.047 -0.158 -0.208 -0.000 -0.051 -0.327 -0.176 -0.231 -0.076 -0.042 -0.067 -0.128 -0.144 OM -0.029 -0.070 -0.090 -0.000 -0.043 -0.068 -0.105 -0.038 -0.038 -0.136 -0.053 -0.087 -0.101 CM -0.099 -0.014 -0.019 -0.000 -0.332 -0.193 -0.004 -0.147 -0.008 -0.032 -0.088 -0.138 -0.057 AHF -0.059 -0.117 -0.148 -0.000 -0.153 -0.093 -0.119 -0.051 -0.019 -0.125 -0.005 -0.008 -0.002 Cheese -0.014 -0.074 -0.096 0.000 -0.005 -0.044 -0.169 -0.396 -0.023 -0.035 -0.021 -0.009 -0.006 Fish -0.108 -0.208 -0.276 -0.002 -0.099 -0.073 -0.259 -0.043 -0.026 -0.103 -0.000 -0.000 -0.000 Yogurt -0.051 -0.063 -0.051 -0.083 -0.058 -0.058 -0.045 -0.032 -0.041 -0.144 -0.025 -0.025 -0.029 PrepM 31 0 Child T Child<15 T <Bac T Bac T >Bac T CSP1 T CSP2 T CSP3 T CSP4 T CSP5 T CSP6 T CSP7 T HomeOw T Juic -0.123 -3.913 0.027 1.297 0.124 7.569 0.057 5.530 0.059 4.936 0.000 -4.201 -0.019 -3.503 -0.037 -2.208 -0.110 -4.883 -0.099 -4.403 -0.093 -4.337 -0.109 -2.916 -0.009 -0.726 Alc 0.027 1.113 -0.032 -1.922 0.086 6.690 0.008 0.959 0.055 5.952 0.000 3.803 0.010 2.465 -0.047 -3.597 -0.041 -2.345 -0.037 -2.112 -0.072 -4.245 -0.037 -1.273 -0.079 -8.125 Soda 0.127 4.357 -0.055 -2.836 -0.002 -0.145 0.014 1.414 -0.028 -2.521 0.000 -2.350 -0.010 -2.059 -0.048 -3.086 -0.146 -6.945 -0.169 -7.930 -0.104 -5.146 -0.272 -7.817 0.067 5.805 Wat -0.144 -9.244 -0.004 -0.376 -0.012 -1.520 0.001 0.210 -0.044 -7.345 0.000 -4.319 -0.015 -5.630 -0.012 -1.418 -0.016 -1.450 -0.015 -1.317 0.030 2.776 -0.023 -1.227 0.033 5.499 Table 5: Socio-economic elasticities Cof 0.065 3.940 -0.056 -5.205 0.062 7.244 0.015 2.788 0.051 8.076 0.000 -1.836 -0.006 -2.237 -0.014 -1.577 -0.012 -1.055 -0.019 -1.565 -0.022 -1.981 0.041 2.119 -0.032 -5.042 FF&V -0.009 -0.387 -0.088 -5.616 -0.068 -5.485 0.014 1.810 0.001 0.160 0.000 0.460 0.000 -0.001 -0.008 -0.607 0.017 0.978 0.011 0.633 0.085 5.238 0.005 0.172 0.047 5.048 Grains -0.046 -0.798 -0.077 -1.982 0.000 -0.015 -0.041 -2.116 -0.030 -1.331 0.000 -1.637 -0.010 -1.030 -0.049 -1.622 -0.126 -3.068 -0.133 -3.214 -0.098 -2.487 -0.124 -1.824 0.047 2.024 VHF -0.036 -1.832 -0.005 -0.407 0.007 0.692 0.002 0.405 -0.023 -3.083 0.000 -2.071 0.002 0.467 -0.052 -4.917 -0.065 -4.521 -0.027 -1.801 -0.062 -4.173 -0.119 -4.854 0.042 5.678 VD 0.192 7.318 0.096 5.458 -0.100 -7.283 -0.022 -2.557 -0.032 -3.267 0.000 -1.978 -0.010 -2.149 -0.002 -0.113 0.039 2.101 0.053 2.814 0.014 0.759 -0.078 -2.503 0.002 0.161 VHS -0.111 -4.541 -0.062 -3.912 0.003 0.227 0.010 1.326 -0.040 -4.278 0.000 2.512 -0.007 -1.658 0.025 1.879 0.012 0.683 -0.033 -1.853 -0.015 -0.869 0.046 1.566 -0.011 -1.135 Starch 0.074 5.402 0.033 3.782 -0.063 -9.008 -0.021 -4.855 -0.030 -5.896 0.000 -0.002 0.003 1.469 0.021 2.805 0.033 3.362 0.025 2.393 0.029 2.862 0.042 2.518 0.018 3.375 PF&V 0.096 7.338 0.037 4.407 0.018 2.753 0.012 2.993 0.020 4.120 0.000 0.806 0.001 0.640 -0.001 -0.154 -0.019 -1.952 0.012 1.162 -0.015 -1.505 -0.037 -2.310 -0.026 -5.332 Beef -0.010 -0.303 0.038 1.758 -0.100 -5.950 -0.052 -4.886 -0.081 -6.654 0.000 -0.841 -0.001 -0.178 -0.016 -1.028 -0.006 -0.286 0.023 1.082 -0.071 -3.419 -0.168 -4.631 0.070 5.466 OM -0.004 -0.230 0.026 2.137 -0.049 -5.157 -0.007 -1.182 -0.075 -10.792 0.000 -1.149 0.000 -0.035 0.011 1.166 0.032 2.437 0.003 0.224 -0.016 -1.244 -0.059 -2.747 0.051 7.200 CM -0.080 -6.496 -0.030 -3.794 -0.006 -0.946 0.009 2.297 0.017 3.571 0.000 0.030 0.007 3.548 0.024 3.686 0.078 8.742 0.049 5.340 0.053 6.024 0.121 8.185 -0.012 -2.433 AHF -0.026 -0.857 -0.052 -2.560 0.021 1.337 0.010 0.952 0.034 2.887 0.000 4.419 0.005 0.906 0.030 1.830 0.039 1.750 0.036 1.634 0.078 3.649 0.103 2.825 -0.013 -1.108 Cheese -0.030 -1.090 0.002 0.101 0.109 7.564 0.046 5.035 0.111 10.537 0.000 2.029 0.010 2.236 0.030 2.061 0.045 2.278 0.039 1.983 0.072 3.856 0.260 8.059 -0.101 -9.263 Fish 0.087 4.193 0.018 1.289 -0.046 -4.224 -0.056 -8.190 -0.042 -5.307 0.000 1.892 0.002 0.580 0.030 2.743 0.053 3.573 0.035 2.320 0.019 1.298 0.028 1.136 0.027 3.267 Yogurt -0.034 -2.775 0.058 7.206 -0.027 -4.166 0.022 5.680 -0.020 -4.262 0.000 0.424 0.005 2.585 0.043 6.478 0.032 3.513 0.021 2.239 0.030 3.335 -0.005 -0.341 0.024 5.137 PrepM 0.163 8.542 0.097 7.741 -0.085 -8.571 -0.055 -8.920 0.018 2.458 0.000 0.680 0.004 1.231 0.005 0.490 0.013 0.980 0.051 3.607 0.127 9.546 0.191 8.425 0.004 0.572 (a) CO2 (b) SO2 (c) N Figure 1: Contribution of Aggregated Food Groups to CO2 , SO2 and Nitrates Emissions 32 Table 6: Food Groups Sample Mean Budget-Shares and log-prices 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Food Groups Labels Juices Alcohol Soft drinks Bottled water Coffee and tea Fresh fruits and vegetables Grains and condiments Plant-based foods high in fats Plant-based dishes Plant-based foods high in sugar Starchy foods Processed fruits and vegetables Beef Other meats (lamb, chicken, pork) Cooked meats Animal-based foods high in fats Cheese Fish and seafoods Yogurts Prepared mixed meals Prepared desserts Juic Alc Soda Wat Cof FF&V Grains VHF VD VHS Starch PF&V Beef OM CM AHF Cheese Fish Yogurt PrepM PrepD Budget shares (wi ) Mean Std. Dev 0.051 0.011 0.100 0.025 0.039 0.011 0.055 0.010 0.046 0.007 0.027 0.005 0.015 0.004 0.027 0.005 0.038 0.009 0.038 0.009 0.023 0.003 0.023 0.003 0.087 0.020 0.059 0.009 0.047 0.006 0.027 0.004 0.079 0.020 0.056 0.013 0.062 0.014 0.049 0.010 0.052 0.008 33 log-price (ln pi ) Mean Std. Dev. -11.235 0.893 -10.752 1.134 0.949 -11.715 -12.814 0.918 -9.795 1.249 -12.432 1.230 -10.221 1.251 1.106 -10.63 1.090 -9.690 -11.213 1.072 -12.101 1.099 1.130 -11.992 -9.175 1.107 -10.670 1.020 -10.462 1.139 -10.834 1.091 -10.524 1.058 -9.831 1.032 0.961 -12.266 -10.897 1.112 -11.306 1.082 Table 7: Proportion of Households for Each Sociodemographic Variable Sociodemographic Variables Mean Std. Dev. 0.501 0.338 0.417 0.153 0.235 0.527 0.333 0.308 0.167 0.084 0.204 0.246 0.012 0.025 0.127 0.185 0.174 0.176 0.261 0.023 0.023 0.149 0.137 0.115 0.159 0.364 0.085 0.205 0.249 0.175 0.285 0.162 0.213 0.175 0.153 0.313 0.720 0.535 0.492 0.130 0.166 0.155 Into the demand system Without child With at least one child (<15) Low degree diploma Level of baccalaureate Baccalaureate and higher degree Home owners Socio-professional category Farmers Craftsmen Executives Intermediary profession Employees Workers Retired Into the unit value equations Income ] ; 900[ Income [900; 1500[ Income [1500; 2300[ Income [2300; 3000[ Income [3000; [ Durable ownership Fryer Dish washer Freezer 34 Table 8: Environmental Emissions, Nutrient Intakes and Percentage of Quantity Change in Total Purchased with 20 % Targeted Taxes Variable Unit Average Household Purchases Daily Equivalent Descriptive Statistics (a) Mean Std. Dev. Percentage of Quantity Change Impact of Tax (%) (b) ENV ENV-NUT Environmental Emissions CO2 g eq. CO2 SO2 g eq. SO2 N g eq. N 3913.817 44.615 15.120 1313.980 14.790 4.769 -9.987 -16.621 -10.286 -8.224 -13.877 -7.297 Nutrient Intakes Energy Proteins Vegetal Proteins Animal Proteins Saturated fats Cholesterol Vitamin B12 Vitamin D Iron Sodium 3081.496 102.641 22.680 78.178 58.885 494.838 7.152 2.498 18.500 3868.042 833.159 26.915 8.927 21.250 15.776 143.36 1.839 0.755 26.876 2304.655 -8.858 -17.034 -2.286 -21.502 -15.596 -18.767 -20.586 -18.767 -6.409 -12.515 -5.499 -8.283 -2.236 -9.941 -12.510 -9.924 -8.586 -5.432 -3.917 -8.956 kcal g g g g mg μg μg mg mg 35 Table 9: Unit Value Equations per Food Group Juic. ln(q) ln(x) Income [900; 1500[ Income [1500; 2300[ Income [2300; 3000[ Income [3000; [ Fryer Dish washer Freezer Cons ln(q) ln(x) Income [900; 1500[ Income [1500; 2300[ Income [2300; 3000[ Income [3000; [ Fryer Dish washer Freezer Cons -0.732‡ (0.005) 0.966‡ (0.026) -0.079‡ (0.028) 0.110‡ (0.033) 0.273‡ (0.044) 0.266‡ (0.048) -0.043* (0.026) -0.085‡ (0.025) -0.165‡ (0.024) -3.410‡ (0.074) Alc. -0.803‡ (0.006) 1.106‡ (0.037) -0.304‡ (0.039) -0.075 (0.047) -0.065 (0.062) -0.089 (0.068) -0.103‡ (0.036) -0.096‡ (0.034) -0.447‡ (0.033) -1.485‡ (0.100) Soda Water -0.701‡ (0.005) 0.905‡ (0.033) 0.202‡ (0.035) 0.070* (0.041) 0.195‡ (0.054) 0.137† (0.060) -0.230‡ (0.032) -0.219‡ (0.031) -0.052* (0.030) -3.918‡ (0.084) -0.780‡ (0.005) 0.955‡ (0.027) -0.118‡ (0.028) -0.138‡ (0.033) -0.157‡ (0.044) -0.219‡ (0.048) 0.086‡ (0.026) 0.139‡ (0.025) 0.003 (0.024) -3.177‡ (0.085) Cof. -1.002‡ (0.005) 0.817‡ (0.026) -0.151‡ (0.029) -0.087† (0.034) -0.079* (0.044) -0.146‡ (0.049) 0.015 (0.026) -0.019 (0.025) -0.282‡ (0.024) 0.787‡ (0.071) F.F&V. Grains V.H.F. -0.909‡ (0.005) 0.768‡ (0.025) -0.071‡ (0.027) -0.131‡ (0.032) -0.223‡ (0.042) -0.280‡ (0.046) 0.198‡ (0.025) -0.055† (0.024) 0.011 (0.023) -1.350‡ (0.075) -0.981‡ (0.005) 0.931‡ (0.037) -0.414‡ (0.040) -0.735‡ (0.047) -0.876‡ (0.062) -0.899‡ (0.068) -0.023 (0.037) 0.356‡ (0.035) -0.041 (0.034) -0.609‡ (0.086) -0.896‡ (0.005) 0.637‡ (0.024) -0.113‡ (0.026) -0.205‡ (0.030) -0.242‡ (0.040) -0.289‡ (0.044) 0.079‡ (0.023) 0.104‡ (0.022) -0.165‡ (0.022) -0.965‡ (0.065) P.F&V. Beef O.M. C.M. A.H.F. Cheese -0.971‡ (0.004) 0.674‡ (0.017) -0.013 (0.019) 0.116‡ (0.022) 0.204‡ (0.029) 0.227‡ (0.032) 0.014 (0.017) -0.096‡ (0.016) -0.146‡ (0.016) -0.407‡ (0.057) -0.865‡ (0.006) 1.148‡ (0.037) 0.032 (0.039) 0.146‡ (0.046) 0.123† (0.061) 0.151† (0.067) -0.125‡ (0.036) 0.110‡ (0.034) -0.043 (0.033) -1.373‡ (0.085) -0.841‡ (0.005) 0.910‡ (0.026) 0.008 (0.027) 0.096‡ (0.032) 0.138‡ (0.042) 0.132‡ (0.047) 0.034 (0.025) 0.112‡ (0.024) -0.049† (0.023) -1.812‡ (0.074) -0.910‡ (0.004) 0.745‡ (0.019) 0.226‡ (0.021) 0.326‡ (0.024) 0.423‡ (0.032) 0.428‡ (0.035) -0.086‡ (0.019) -0.079‡ (0.018) -0.036† (0.018) -0.868‡ (0.054) -0.894‡ (0.004) 0.746‡ (0.018) -0.047† (0.019) -0.046† (0.023) -0.030 (0.030) 0.005 (0.033) 0.053‡ (0.018) 0.052‡ (0.017) -0.011 (0.017) -1.387‡ (0.056) -0.826‡ (0.006) 0.972‡ (0.023) 0.139‡ (0.025) 0.283‡ (0.029) 0.410‡ (0.039) 0.401‡ (0.043) -0.005 (0.023) 0.061‡ (0.022) -0.260‡ (0.021) -2.125‡ (0.079) Fish -0.876‡ (0.006) 0.889‡ (0.028) -0.124‡ (0.030) -0.065* (0.035) -0.110† (0.047) -0.101† (0.051) 0.029 (0.027) 0.095‡ (0.026) -0.173‡ (0.025) -0.840‡ (0.077) Yogurt -0.836‡ (0.004) 0.867‡ (0.019) 0.172‡ (0.020) 0.207‡ (0.024) 0.301‡ (0.032) 0.363‡ (0.035) -0.002 (0.019) 0.089‡ (0.018) 0.092‡ (0.018) -2.222‡ (0.064) V. dishes -0.892‡ (0.005) 0.933‡ (0.036) -0.015 (0.039) 0.146‡ (0.046) 0.150† (0.061) 0.052 (0.067) 0.066* (0.036) -0.105‡ (0.034) -0.084† (0.033) -1.335‡ (0.085) V.H.S. Starch. -0.886‡ (0.006) 0.694‡ (0.028) -0.018 (0.030) 0.152‡ (0.035) 0.311‡ (0.047) 0.256‡ (0.051) -0.265‡ (0.027) 0.071‡ (0.026) -0.002 (0.026) -0.748‡ (0.085) -0.937‡ (0.004) 0.615‡ (0.017) -0.013 (0.018) 0.026 (0.021) 0.073‡ (0.028) 0.059* (0.030) 0.005 (0.016) -0.019 (0.016) -0.006 (0.015) -0.821‡ (0.054) Prep. meals Prep. des. -0.992‡ (0.006) 0.949‡ (0.029) 0.061† (0.031) -0.190‡ (0.037) -0.353‡ (0.048) -0.413‡ (0.053) 0.312‡ (0.028) 0.229‡ (0.027) 0.087‡ (0.026) -0.124 (0.090) -0.894‡ (0.004) 0.771‡ (0.018) 0.071‡ (0.019) 0.174‡ (0.023) 0.314‡ (0.030) 0.256‡ (0.033) -0.019 (0.018) 0.007 (0.017) -0.026 (0.017) -1.111‡ (0.060) Notes: Robust standard errors are bellow coefficients. Significance levels: ‡=1% ; †=5% ; *=10%. Juic.=Juices; Alc.=Alcohol; Soda=Soft drinks; Water=Botteld water; Cof.=Coffee/tea; F. F&V=Fresh fruits and vegetables; V.H.F.=Plant-based foods high in fats; V. dishes=Plant-based foods dishes; V.H.S.=Plant-based foods high in sugar; Starch.=Starchy foods; P. F&V=Processed fruits and vegetables; O. M.=Other meats, e.g. Lamb poultry, pork, etc.; C. M.=Cooked meats, e.g. ham; A.H.F.=Animal-based foods with high fat content; Cheese=Cheese; Fish=Fish and seafoods; Yogurts; Prep. meals=prepared meals; Prep. des=prepared desserts. Year, four-week period dummies and year dummies are not reported in interests of space. They are available upon request. 36 Table 10: Percentage of Quantity Change in Total Purchased with 20 % Targeted Taxes ENVNUT Well-Off ENV CO2 SO2 N Energy Proteins Vegetal Proteins Animal Proteins Saturated Fats Cholesterol Vitamin B12 Vitamin D Iron Sodium -9.085 -15.791 -9.594 -8.365 -16.198 -2.075 -20.745 -14.536 -18.305 -19.582 -17.802 -6.492 -12.250 -7.565 -13.377 -6.970 -5.067 -7.884 -2.019 -9.509 -11.606 -8.979 -7.553 -4.452 -3.854 -8.576 CO2 SO2 N Energy Proteins Vegetal Proteins Animal Proteins Saturated Fats Cholesterol Vitamin B12 Vitamin D Iron Sodium -10.003 -16.608 -10.629 -9.035 -17.034 -2.039 -21.472 -15.463 -18.711 -20.751 -18.372 -6.557 -12.513 -8.256 -13.957 -7.648 -5.576 -8.528 -1.978 -10.227 -12.463 -9.741 -8.640 -4.938 -3.949 -8.948 CO2 SO2 N Energy Proteins Vegetal Proteins Animal Proteins Saturated Fats Cholesterol Vitamin B12 Vitamin D Iron Sodium -9.828 -16.429 -10.308 -8.793 -17.010 -2.034 -21.328 -15.772 -18.936 -20.702 -19.141 -6.195 -12.901 -7.976 -13.524 -7.310 -5.757 -8.980 -1.972 -10.822 -13.042 -10.479 -9.578 -5.630 -3.843 -9.732 CO2 SO2 N Energy Proteins Vegetal Proteins Animal Proteins Saturated Fats Cholesterol Vitamin B12 Vitamin D Iron Sodium -10.066 -16.579 -10.377 -8.978 -17.092 -2.055 -21.277 -16.424 -19.116 -20.905 -19.585 -5.793 -12.939 -8.242 -13.721 -7.531 -5.978 -9.166 -1.996 -10.991 -13.742 -11.489 -9.997 -6.410 -3.690 -9.890 ENVENVENV NUT NUT Upper-Average Lower-Average Age less than 30 -9.848 -8.225 -10.318 -8.612 -16.532 -14.059 -16.959 -14.386 -10.173 -7.390 -10.470 -7.493 -9.142 -5.269 -9.506 -5.271 -17.710 -7.863 -18.074 -7.487 -2.346 -2.291 -2.488 -2.436 -22.127 -9.320 -22.490 -8.811 -15.054 -11.843 -15.374 -11.827 -18.532 -9.235 -18.802 -9.272 -21.233 -7.484 -21.530 -7.100 -18.357 -4.671 -18.553 -4.951 -6.802 -4.038 -7.023 -4.119 -13.073 -8.594 -13.822 -8.431 Age between 30 and 45 -10.001 -8.199 -10.290 -8.472 -16.670 -13.879 -16.952 -14.162 -10.496 -7.397 -10.601 -7.454 -9.325 -5.488 -9.489 -5.551 -17.530 -7.981 -17.788 -8.018 -2.135 -2.080 -2.327 -2.273 -22.005 -9.564 -22.283 -9.553 -15.467 -12.180 -15.762 -12.241 -18.602 -9.460 -18.764 -9.443 -21.178 -8.024 -21.482 -8.075 -18.227 -5.077 -18.440 -5.278 -6.627 -3.943 -6.796 -4.042 -13.269 -8.849 -13.542 -8.969 Age between 45 and 60 -9.885 -8.084 -9.804 -8.048 -16.550 -13.732 -16.524 -13.733 -10.364 -7.303 -10.197 -7.142 -8.918 -5.712 -8.889 -5.619 -17.196 -8.779 -17.316 -8.513 -2.190 -2.132 -2.241 -2.188 -21.669 -10.576 -21.949 -10.227 -15.779 -12.846 -16.109 -13.013 -18.776 -10.242 -19.091 -10.201 -20.914 -9.451 -20.990 -9.133 -18.873 -5.612 -19.009 -5.689 -6.350 -3.919 -6.317 -3.879 -13.132 -9.628 -13.324 -9.599 Age more than 60 -10.044 -8.215 -10.174 -8.399 -16.668 -13.795 -16.857 -14.061 -10.315 -7.355 -10.260 -7.327 -8.833 -5.810 -8.729 -5.712 -17.091 -8.923 -17.180 -8.765 -2.068 -2.011 -2.220 -2.171 -21.570 -10.744 -21.679 -10.535 -16.261 -13.528 -16.426 -13.636 -19.183 -11.207 -19.417 -11.524 37 -21.073 -9.963 -21.178 -10.079 -19.730 -6.315 -19.907 -6.605 -5.820 -3.684 -5.862 -3.743 -13.236 -9.928 -13.155 -9.716 ENV ENV ENVNUT Modest -10.316 -16.842 -10.276 -9.169 -17.646 -2.744 -22.135 -15.366 -19.002 -21.289 -18.386 -7.191 -13.299 -8.581 -14.231 -7.344 -5.220 -7.651 -2.701 -8.973 -11.633 -9.042 -7.202 -4.667 -4.271 -8.659 -10.308 -16.907 -10.391 -9.399 -17.938 -2.612 -22.470 -15.732 -19.062 -21.496 -18.678 -6.925 -13.573 -8.501 -14.101 -7.280 -5.405 -7.787 -2.565 -9.238 -11.950 -9.236 -7.735 -5.173 -4.101 -8.935 -9.713 -16.442 -9.812 -8.730 -17.286 -2.348 -22.043 -16.188 -19.135 -21.118 -18.905 -6.220 -13.006 -8.065 -13.795 -6.885 -5.485 -8.199 -2.306 -9.917 -12.969 -10.189 -8.746 -5.897 -3.821 -9.297 -10.059 -16.698 -10.053 -8.600 -17.285 -2.277 -21.811 -16.399 -19.345 -21.078 -19.784 -6.074 -13.114 -8.241 -13.857 -7.029 -5.487 -8.339 -2.236 -9.960 -13.374 -11.009 -9.353 -6.409 -3.777 -9.715